blk-merge.c 20.9 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2 3 4 5 6 7 8 9 10
/*
 * 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>

11 12
#include <trace/events/block.h>

13 14
#include "blk.h"

15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
static inline bool bio_will_gap(struct request_queue *q,
		struct request *prev_rq, struct bio *prev, struct bio *next)
{
	struct bio_vec pb, nb;

	if (!bio_has_data(prev) || !queue_virt_boundary(q))
		return false;

	/*
	 * Don't merge if the 1st bio starts with non-zero offset, otherwise it
	 * is quite difficult to respect the sg gap limit.  We work hard to
	 * merge a huge number of small single bios in case of mkfs.
	 */
	if (prev_rq)
		bio_get_first_bvec(prev_rq->bio, &pb);
	else
		bio_get_first_bvec(prev, &pb);
32
	if (pb.bv_offset & queue_virt_boundary(q))
33 34 35 36 37 38 39 40 41 42 43 44 45
		return true;

	/*
	 * We don't need to worry about the situation that the merged segment
	 * ends in unaligned virt boundary:
	 *
	 * - if 'pb' ends aligned, the merged segment ends aligned
	 * - if 'pb' ends unaligned, the next bio must include
	 *   one single bvec of 'nb', otherwise the 'nb' can't
	 *   merge with 'pb'
	 */
	bio_get_last_bvec(prev, &pb);
	bio_get_first_bvec(next, &nb);
46
	if (biovec_phys_mergeable(q, &pb, &nb))
47 48 49 50 51 52 53 54 55 56 57 58 59 60
		return false;
	return __bvec_gap_to_prev(q, &pb, nb.bv_offset);
}

static inline bool req_gap_back_merge(struct request *req, struct bio *bio)
{
	return bio_will_gap(req->q, req, req->biotail, bio);
}

static inline bool req_gap_front_merge(struct request *req, struct bio *bio)
{
	return bio_will_gap(req->q, NULL, bio, req->bio);
}

61 62
static struct bio *blk_bio_discard_split(struct request_queue *q,
					 struct bio *bio,
63 64
					 struct bio_set *bs,
					 unsigned *nsegs)
65 66 67 68 69 70
{
	unsigned int max_discard_sectors, granularity;
	int alignment;
	sector_t tmp;
	unsigned split_sectors;

71 72
	*nsegs = 1;

73 74 75
	/* Zero-sector (unknown) and one-sector granularities are the same.  */
	granularity = max(q->limits.discard_granularity >> 9, 1U);

76 77
	max_discard_sectors = min(q->limits.max_discard_sectors,
			bio_allowed_max_sectors(q));
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
	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);
}

105 106 107 108 109 110 111 112 113 114 115 116 117 118
static struct bio *blk_bio_write_zeroes_split(struct request_queue *q,
		struct bio *bio, struct bio_set *bs, unsigned *nsegs)
{
	*nsegs = 1;

	if (!q->limits.max_write_zeroes_sectors)
		return NULL;

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

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

119 120
static struct bio *blk_bio_write_same_split(struct request_queue *q,
					    struct bio *bio,
121 122
					    struct bio_set *bs,
					    unsigned *nsegs)
123
{
124 125
	*nsegs = 1;

126 127 128 129 130 131 132 133 134
	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);
}

135 136 137 138 139 140 141 142 143 144 145 146
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;
}

147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164
static unsigned get_max_segment_size(struct request_queue *q,
				     unsigned offset)
{
	unsigned long mask = queue_segment_boundary(q);

	/* default segment boundary mask means no boundary limit */
	if (mask == BLK_SEG_BOUNDARY_MASK)
		return queue_max_segment_size(q);

	return min_t(unsigned long, mask - (mask & offset) + 1,
		     queue_max_segment_size(q));
}

/*
 * Split the bvec @bv into segments, and update all kinds of
 * variables.
 */
static bool bvec_split_segs(struct request_queue *q, struct bio_vec *bv,
165
		unsigned *nsegs, unsigned *sectors, unsigned max_segs)
166 167 168 169 170 171 172 173 174
{
	unsigned len = bv->bv_len;
	unsigned total_len = 0;
	unsigned new_nsegs = 0, seg_size = 0;

	/*
	 * Multi-page bvec may be too big to hold in one segment, so the
	 * current bvec has to be splitted as multiple segments.
	 */
175
	while (len && new_nsegs + *nsegs < max_segs) {
176 177 178 179 180 181 182 183 184 185 186
		seg_size = get_max_segment_size(q, bv->bv_offset + total_len);
		seg_size = min(seg_size, len);

		new_nsegs++;
		total_len += seg_size;
		len -= seg_size;

		if ((bv->bv_offset + total_len) & queue_virt_boundary(q))
			break;
	}

187 188 189 190
	if (new_nsegs) {
		*nsegs += new_nsegs;
		if (sectors)
			*sectors += total_len >> 9;
191 192 193 194 195 196
	}

	/* split in the middle of the bvec if len != 0 */
	return !!len;
}

197 198
static struct bio *blk_bio_segment_split(struct request_queue *q,
					 struct bio *bio,
199 200
					 struct bio_set *bs,
					 unsigned *segs)
201
{
202
	struct bio_vec bv, bvprv, *bvprvp = NULL;
203
	struct bvec_iter iter;
204
	unsigned nsegs = 0, sectors = 0;
M
Ming Lei 已提交
205 206
	bool do_split = true;
	struct bio *new = NULL;
207
	const unsigned max_sectors = get_max_io_size(q, bio);
208
	const unsigned max_segs = queue_max_segments(q);
209

210
	bio_for_each_bvec(bv, bio, iter) {
211 212 213 214
		/*
		 * If the queue doesn't support SG gaps and adding this
		 * offset would create a gap, disallow it.
		 */
215
		if (bvprvp && bvec_gap_to_prev(q, bvprvp, bv.bv_offset))
216 217
			goto split;

218
		if (sectors + (bv.bv_len >> 9) > max_sectors) {
219 220 221 222
			/*
			 * Consider this a new segment if we're splitting in
			 * the middle of this vector.
			 */
223
			if (nsegs < max_segs &&
224
			    sectors < max_sectors) {
225 226 227
				/* split in the middle of bvec */
				bv.bv_len = (max_sectors - sectors) << 9;
				bvec_split_segs(q, &bv, &nsegs,
228
						&sectors, max_segs);
229
			}
230
			goto split;
231 232
		}

233
		if (nsegs == max_segs)
234 235 236
			goto split;

		bvprv = bv;
M
Ming Lei 已提交
237
		bvprvp = &bvprv;
238

239 240 241
		if (bv.bv_offset + bv.bv_len <= PAGE_SIZE) {
			nsegs++;
			sectors += bv.bv_len >> 9;
242 243
		} else if (bvec_split_segs(q, &bv, &nsegs, &sectors,
				max_segs)) {
244
			goto split;
245
		}
246 247
	}

M
Ming Lei 已提交
248
	do_split = false;
249
split:
250
	*segs = nsegs;
M
Ming Lei 已提交
251 252 253 254 255 256 257 258

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

	return do_split ? new : NULL;
259 260
}

261 262
void __blk_queue_split(struct request_queue *q, struct bio **bio,
		unsigned int *nr_segs)
263
{
264
	struct bio *split;
265

A
Adrian Hunter 已提交
266 267 268
	switch (bio_op(*bio)) {
	case REQ_OP_DISCARD:
	case REQ_OP_SECURE_ERASE:
269
		split = blk_bio_discard_split(q, *bio, &q->bio_split, nr_segs);
A
Adrian Hunter 已提交
270
		break;
271
	case REQ_OP_WRITE_ZEROES:
272 273
		split = blk_bio_write_zeroes_split(q, *bio, &q->bio_split,
				nr_segs);
274
		break;
A
Adrian Hunter 已提交
275
	case REQ_OP_WRITE_SAME:
276 277
		split = blk_bio_write_same_split(q, *bio, &q->bio_split,
				nr_segs);
A
Adrian Hunter 已提交
278 279
		break;
	default:
280
		split = blk_bio_segment_split(q, *bio, &q->bio_split, nr_segs);
A
Adrian Hunter 已提交
281 282
		break;
	}
283

284
	if (split) {
M
Ming Lei 已提交
285
		/* there isn't chance to merge the splitted bio */
J
Jens Axboe 已提交
286
		split->bi_opf |= REQ_NOMERGE;
M
Ming Lei 已提交
287

288 289 290 291 292 293 294 295 296 297
		/*
		 * Since we're recursing into make_request here, ensure
		 * that we mark this bio as already having entered the queue.
		 * If not, and the queue is going away, we can get stuck
		 * forever on waiting for the queue reference to drop. But
		 * that will never happen, as we're already holding a
		 * reference to it.
		 */
		bio_set_flag(*bio, BIO_QUEUE_ENTERED);

298
		bio_chain(split, *bio);
299
		trace_block_split(q, split, (*bio)->bi_iter.bi_sector);
300 301 302 303
		generic_make_request(*bio);
		*bio = split;
	}
}
304 305 306 307 308 309 310

void blk_queue_split(struct request_queue *q, struct bio **bio)
{
	unsigned int nr_segs;

	__blk_queue_split(q, bio, &nr_segs);
}
311 312
EXPORT_SYMBOL(blk_queue_split);

313
static unsigned int __blk_recalc_rq_segments(struct request_queue *q,
M
Ming Lei 已提交
314
					     struct bio *bio)
315
{
316
	unsigned int nr_phys_segs = 0;
317
	struct bvec_iter iter;
318
	struct bio_vec bv;
319

320 321
	if (!bio)
		return 0;
322

323 324 325 326
	switch (bio_op(bio)) {
	case REQ_OP_DISCARD:
	case REQ_OP_SECURE_ERASE:
	case REQ_OP_WRITE_ZEROES:
327 328
		return 0;
	case REQ_OP_WRITE_SAME:
329
		return 1;
330
	}
331

332
	for_each_bio(bio) {
333 334
		bio_for_each_bvec(bv, bio, iter)
			bvec_split_segs(q, &bv, &nr_phys_segs, NULL, UINT_MAX);
335 336
	}

337 338 339 340 341
	return nr_phys_segs;
}

void blk_recalc_rq_segments(struct request *rq)
{
M
Ming Lei 已提交
342
	rq->nr_phys_segments = __blk_recalc_rq_segments(rq->q, rq->bio);
343 344
}

345
static inline struct scatterlist *blk_next_sg(struct scatterlist **sg,
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
		struct scatterlist *sglist)
{
	if (!*sg)
		return sglist;

	/*
	 * 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.
	 */
	sg_unmark_end(*sg);
	return sg_next(*sg);
}

static unsigned blk_bvec_map_sg(struct request_queue *q,
		struct bio_vec *bvec, struct scatterlist *sglist,
		struct scatterlist **sg)
{
	unsigned nbytes = bvec->bv_len;
367
	unsigned nsegs = 0, total = 0;
368 369

	while (nbytes > 0) {
370 371
		unsigned offset = bvec->bv_offset + total;
		unsigned len = min(get_max_segment_size(q, offset), nbytes);
372 373 374 375 376 377 378 379 380 381 382 383
		struct page *page = bvec->bv_page;

		/*
		 * Unfortunately a fair number of drivers barf on scatterlists
		 * that have an offset larger than PAGE_SIZE, despite other
		 * subsystems dealing with that invariant just fine.  For now
		 * stick to the legacy format where we never present those from
		 * the block layer, but the code below should be removed once
		 * these offenders (mostly MMC/SD drivers) are fixed.
		 */
		page += (offset >> PAGE_SHIFT);
		offset &= ~PAGE_MASK;
384 385

		*sg = blk_next_sg(sg, sglist);
386
		sg_set_page(*sg, page, len, offset);
387

388 389
		total += len;
		nbytes -= len;
390 391 392 393 394 395
		nsegs++;
	}

	return nsegs;
}

396 397 398 399 400 401 402 403
static inline int __blk_bvec_map_sg(struct bio_vec bv,
		struct scatterlist *sglist, struct scatterlist **sg)
{
	*sg = blk_next_sg(sg, sglist);
	sg_set_page(*sg, bv.bv_page, bv.bv_len, bv.bv_offset);
	return 1;
}

404 405 406 407
/* only try to merge bvecs into one sg if they are from two bios */
static inline bool
__blk_segment_map_sg_merge(struct request_queue *q, struct bio_vec *bvec,
			   struct bio_vec *bvprv, struct scatterlist **sg)
408 409 410 411
{

	int nbytes = bvec->bv_len;

412 413
	if (!*sg)
		return false;
414

415 416 417 418 419 420 421 422 423
	if ((*sg)->length + nbytes > queue_max_segment_size(q))
		return false;

	if (!biovec_phys_mergeable(q, bvprv, bvec))
		return false;

	(*sg)->length += nbytes;

	return true;
424 425
}

426 427 428
static int __blk_bios_map_sg(struct request_queue *q, struct bio *bio,
			     struct scatterlist *sglist,
			     struct scatterlist **sg)
429
{
430
	struct bio_vec uninitialized_var(bvec), bvprv = { NULL };
431
	struct bvec_iter iter;
432
	int nsegs = 0;
433
	bool new_bio = false;
434

435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452
	for_each_bio(bio) {
		bio_for_each_bvec(bvec, bio, iter) {
			/*
			 * Only try to merge bvecs from two bios given we
			 * have done bio internal merge when adding pages
			 * to bio
			 */
			if (new_bio &&
			    __blk_segment_map_sg_merge(q, &bvec, &bvprv, sg))
				goto next_bvec;

			if (bvec.bv_offset + bvec.bv_len <= PAGE_SIZE)
				nsegs += __blk_bvec_map_sg(bvec, sglist, sg);
			else
				nsegs += blk_bvec_map_sg(q, &bvec, sglist, sg);
 next_bvec:
			new_bio = false;
		}
453 454 455 456
		if (likely(bio->bi_iter.bi_size)) {
			bvprv = bvec;
			new_bio = true;
		}
457
	}
458

459 460 461 462 463 464 465 466 467 468 469 470 471
	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;

472
	if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
473
		nsegs = __blk_bvec_map_sg(rq->special_vec, sglist, &sg);
474
	else if (rq->bio && bio_op(rq->bio) == REQ_OP_WRITE_SAME)
475
		nsegs = __blk_bvec_map_sg(bio_iovec(rq->bio), sglist, &sg);
476
	else if (rq->bio)
477
		nsegs = __blk_bios_map_sg(q, rq->bio, sglist, &sg);
478

479
	if (unlikely(rq->rq_flags & RQF_COPY_USER) &&
480 481 482
	    (blk_rq_bytes(rq) & q->dma_pad_mask)) {
		unsigned int pad_len =
			(q->dma_pad_mask & ~blk_rq_bytes(rq)) + 1;
483 484 485 486 487

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

488
	if (q->dma_drain_size && q->dma_drain_needed(rq)) {
489
		if (op_is_write(req_op(rq)))
490 491
			memset(q->dma_drain_buffer, 0, q->dma_drain_size);

492
		sg_unmark_end(sg);
493 494 495 496 497 498
		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++;
499
		rq->extra_len += q->dma_drain_size;
500 501 502 503 504
	}

	if (sg)
		sg_mark_end(sg);

505 506 507 508
	/*
	 * Something must have been wrong if the figured number of
	 * segment is bigger than number of req's physical segments
	 */
509
	WARN_ON(nsegs > blk_rq_nr_phys_segments(rq));
510

511 512 513 514
	return nsegs;
}
EXPORT_SYMBOL(blk_rq_map_sg);

515 516
static inline int ll_new_hw_segment(struct request *req, struct bio *bio,
		unsigned int nr_phys_segs)
517
{
518
	if (req->nr_phys_segments + nr_phys_segs > queue_max_segments(req->q))
519 520
		goto no_merge;

521
	if (blk_integrity_merge_bio(req->q, req, bio) == false)
522
		goto no_merge;
523 524 525 526 527 528 529

	/*
	 * This will form the start of a new hw segment.  Bump both
	 * counters.
	 */
	req->nr_phys_segments += nr_phys_segs;
	return 1;
530 531

no_merge:
532
	req_set_nomerge(req->q, req);
533
	return 0;
534 535
}

536
int ll_back_merge_fn(struct request *req, struct bio *bio, unsigned int nr_segs)
537
{
538 539
	if (req_gap_back_merge(req, bio))
		return 0;
540 541 542
	if (blk_integrity_rq(req) &&
	    integrity_req_gap_back_merge(req, bio))
		return 0;
543
	if (blk_rq_sectors(req) + bio_sectors(bio) >
D
Damien Le Moal 已提交
544
	    blk_rq_get_max_sectors(req, blk_rq_pos(req))) {
545
		req_set_nomerge(req->q, req);
546 547 548
		return 0;
	}

549
	return ll_new_hw_segment(req, bio, nr_segs);
550 551
}

552
int ll_front_merge_fn(struct request *req, struct bio *bio, unsigned int nr_segs)
553
{
554 555
	if (req_gap_front_merge(req, bio))
		return 0;
556 557 558
	if (blk_integrity_rq(req) &&
	    integrity_req_gap_front_merge(req, bio))
		return 0;
559
	if (blk_rq_sectors(req) + bio_sectors(bio) >
D
Damien Le Moal 已提交
560
	    blk_rq_get_max_sectors(req, bio->bi_iter.bi_sector)) {
561
		req_set_nomerge(req->q, req);
562 563 564
		return 0;
	}

565
	return ll_new_hw_segment(req, bio, nr_segs);
566 567
}

568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585
static bool req_attempt_discard_merge(struct request_queue *q, struct request *req,
		struct request *next)
{
	unsigned short segments = blk_rq_nr_discard_segments(req);

	if (segments >= queue_max_discard_segments(q))
		goto no_merge;
	if (blk_rq_sectors(req) + bio_sectors(next->bio) >
	    blk_rq_get_max_sectors(req, blk_rq_pos(req)))
		goto no_merge;

	req->nr_phys_segments = segments + blk_rq_nr_discard_segments(next);
	return true;
no_merge:
	req_set_nomerge(q, req);
	return false;
}

586 587 588 589 590
static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
				struct request *next)
{
	int total_phys_segments;

591
	if (req_gap_back_merge(req, next->bio))
592 593
		return 0;

594 595 596
	/*
	 * Will it become too large?
	 */
597
	if ((blk_rq_sectors(req) + blk_rq_sectors(next)) >
D
Damien Le Moal 已提交
598
	    blk_rq_get_max_sectors(req, blk_rq_pos(req)))
599 600 601
		return 0;

	total_phys_segments = req->nr_phys_segments + next->nr_phys_segments;
602
	if (total_phys_segments > queue_max_segments(q))
603 604
		return 0;

605
	if (blk_integrity_merge_rq(q, req, next) == false)
606 607
		return 0;

608 609 610 611 612
	/* Merge is OK... */
	req->nr_phys_segments = total_phys_segments;
	return 1;
}

613 614 615 616 617 618 619 620 621 622 623 624 625 626
/**
 * 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;

627
	if (rq->rq_flags & RQF_MIXED_MERGE)
628 629 630 631 632 633 634 635
		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) {
J
Jens Axboe 已提交
636 637 638
		WARN_ON_ONCE((bio->bi_opf & REQ_FAILFAST_MASK) &&
			     (bio->bi_opf & REQ_FAILFAST_MASK) != ff);
		bio->bi_opf |= ff;
639
	}
640
	rq->rq_flags |= RQF_MIXED_MERGE;
641 642
}

643 644 645 646 647
static void blk_account_io_merge(struct request *req)
{
	if (blk_do_io_stat(req)) {
		struct hd_struct *part;

648
		part_stat_lock();
649
		part = req->part;
650

651
		part_dec_in_flight(req->q, part, rq_data_dir(req));
652

653
		hd_struct_put(part);
654 655 656
		part_stat_unlock();
	}
}
657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672
/*
 * Two cases of handling DISCARD merge:
 * If max_discard_segments > 1, the driver takes every bio
 * as a range and send them to controller together. The ranges
 * needn't to be contiguous.
 * Otherwise, the bios/requests will be handled as same as
 * others which should be contiguous.
 */
static inline bool blk_discard_mergable(struct request *req)
{
	if (req_op(req) == REQ_OP_DISCARD &&
	    queue_max_discard_segments(req->q) > 1)
		return true;
	return false;
}

673 674
static enum elv_merge blk_try_req_merge(struct request *req,
					struct request *next)
675 676 677 678 679 680 681 682
{
	if (blk_discard_mergable(req))
		return ELEVATOR_DISCARD_MERGE;
	else if (blk_rq_pos(req) + blk_rq_sectors(req) == blk_rq_pos(next))
		return ELEVATOR_BACK_MERGE;

	return ELEVATOR_NO_MERGE;
}
683

684
/*
685 686
 * For non-mq, this has to be called with the request spinlock acquired.
 * For mq with scheduling, the appropriate queue wide lock should be held.
687
 */
688 689
static struct request *attempt_merge(struct request_queue *q,
				     struct request *req, struct request *next)
690 691
{
	if (!rq_mergeable(req) || !rq_mergeable(next))
692
		return NULL;
693

694
	if (req_op(req) != req_op(next))
695
		return NULL;
696

697
	if (rq_data_dir(req) != rq_data_dir(next)
698
	    || req->rq_disk != next->rq_disk)
699
		return NULL;
700

701
	if (req_op(req) == REQ_OP_WRITE_SAME &&
702
	    !blk_write_same_mergeable(req->bio, next->bio))
703
		return NULL;
704

705 706 707 708 709 710 711
	/*
	 * Don't allow merge of different write hints, or for a hint with
	 * non-hint IO.
	 */
	if (req->write_hint != next->write_hint)
		return NULL;

712 713 714
	if (req->ioprio != next->ioprio)
		return NULL;

715 716 717 718
	/*
	 * 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
719 720
	 * counts here. Handle DISCARDs separately, as they
	 * have separate settings.
721
	 */
722 723 724

	switch (blk_try_req_merge(req, next)) {
	case ELEVATOR_DISCARD_MERGE:
725 726
		if (!req_attempt_discard_merge(q, req, next))
			return NULL;
727 728 729 730 731 732
		break;
	case ELEVATOR_BACK_MERGE:
		if (!ll_merge_requests_fn(q, req, next))
			return NULL;
		break;
	default:
733
		return NULL;
734
	}
735

736 737 738 739 740 741
	/*
	 * 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.
	 */
742
	if (((req->rq_flags | next->rq_flags) & RQF_MIXED_MERGE) ||
743 744 745 746 747 748
	    (req->cmd_flags & REQ_FAILFAST_MASK) !=
	    (next->cmd_flags & REQ_FAILFAST_MASK)) {
		blk_rq_set_mixed_merge(req);
		blk_rq_set_mixed_merge(next);
	}

749
	/*
750 751 752
	 * At this point we have either done a back merge or front merge. We
	 * need the smaller start_time_ns of the merged requests to be the
	 * current request for accounting purposes.
753
	 */
754 755
	if (next->start_time_ns < req->start_time_ns)
		req->start_time_ns = next->start_time_ns;
756 757 758 759

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

760
	req->__data_len += blk_rq_bytes(next);
761

M
Ming Lei 已提交
762
	if (!blk_discard_mergable(req))
763
		elv_merge_requests(q, req, next);
764

765 766 767 768
	/*
	 * 'next' is going away, so update stats accordingly
	 */
	blk_account_io_merge(next);
769

770 771 772 773
	/*
	 * ownership of bio passed from next to req, return 'next' for
	 * the caller to free
	 */
774
	next->bio = NULL;
775
	return next;
776 777
}

778
struct request *attempt_back_merge(struct request_queue *q, struct request *rq)
779 780 781 782 783 784
{
	struct request *next = elv_latter_request(q, rq);

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

785
	return NULL;
786 787
}

788
struct request *attempt_front_merge(struct request_queue *q, struct request *rq)
789 790 791 792 793 794
{
	struct request *prev = elv_former_request(q, rq);

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

795
	return NULL;
796
}
797 798 799 800

int blk_attempt_req_merge(struct request_queue *q, struct request *rq,
			  struct request *next)
{
801
	struct request *free;
802

803 804
	free = attempt_merge(q, rq, next);
	if (free) {
J
Jens Axboe 已提交
805
		blk_put_request(free);
806 807 808 809
		return 1;
	}

	return 0;
810
}
811 812 813

bool blk_rq_merge_ok(struct request *rq, struct bio *bio)
{
814
	if (!rq_mergeable(rq) || !bio_mergeable(bio))
815 816
		return false;

817
	if (req_op(rq) != bio_op(bio))
818 819
		return false;

820 821 822 823
	/* different data direction or already started, don't merge */
	if (bio_data_dir(bio) != rq_data_dir(rq))
		return false;

824 825
	/* must be same device */
	if (rq->rq_disk != bio->bi_disk)
826 827 828
		return false;

	/* only merge integrity protected bio into ditto rq */
829
	if (blk_integrity_merge_bio(rq->q, rq, bio) == false)
830 831
		return false;

832
	/* must be using the same buffer */
833
	if (req_op(rq) == REQ_OP_WRITE_SAME &&
834 835 836
	    !blk_write_same_mergeable(rq->bio, bio))
		return false;

837 838 839 840 841 842 843
	/*
	 * Don't allow merge of different write hints, or for a hint with
	 * non-hint IO.
	 */
	if (rq->write_hint != bio->bi_write_hint)
		return false;

844 845 846
	if (rq->ioprio != bio_prio(bio))
		return false;

847 848 849
	return true;
}

850
enum elv_merge blk_try_merge(struct request *rq, struct bio *bio)
851
{
852
	if (blk_discard_mergable(rq))
853 854
		return ELEVATOR_DISCARD_MERGE;
	else if (blk_rq_pos(rq) + blk_rq_sectors(rq) == bio->bi_iter.bi_sector)
855
		return ELEVATOR_BACK_MERGE;
856
	else if (blk_rq_pos(rq) - bio_sectors(bio) == bio->bi_iter.bi_sector)
857 858 859
		return ELEVATOR_FRONT_MERGE;
	return ELEVATOR_NO_MERGE;
}