bio.c 39.1 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2
 * Copyright (C) 2001 Jens Axboe <axboe@kernel.dk>
L
Linus Torvalds 已提交
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
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public Licens
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-
 *
 */
#include <linux/mm.h>
#include <linux/swap.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/mempool.h>
#include <linux/workqueue.h>
28
#include <linux/blktrace_api.h>
29
#include <trace/block.h>
30
#include <scsi/sg.h>		/* for struct sg_iovec */
L
Linus Torvalds 已提交
31

32 33
DEFINE_TRACE(block_split);

34 35 36 37 38 39
/*
 * Test patch to inline a certain number of bi_io_vec's inside the bio
 * itself, to shrink a bio data allocation from two mempool calls to one
 */
#define BIO_INLINE_VECS		4

D
Denis ChengRq 已提交
40
static mempool_t *bio_split_pool __read_mostly;
L
Linus Torvalds 已提交
41 42 43 44 45 46 47

/*
 * if you change this list, also change bvec_alloc or things will
 * break badly! cannot be bigger than what you can fit into an
 * unsigned short
 */
#define BV(x) { .nr_vecs = x, .name = "biovec-"__stringify(x) }
48
struct biovec_slab bvec_slabs[BIOVEC_NR_POOLS] __read_mostly = {
L
Linus Torvalds 已提交
49 50 51 52 53 54 55 56
	BV(1), BV(4), BV(16), BV(64), BV(128), BV(BIO_MAX_PAGES),
};
#undef BV

/*
 * fs_bio_set is the bio_set containing bio and iovec memory pools used by
 * IO code that does not need private memory pools.
 */
57
struct bio_set *fs_bio_set;
L
Linus Torvalds 已提交
58

59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153
/*
 * Our slab pool management
 */
struct bio_slab {
	struct kmem_cache *slab;
	unsigned int slab_ref;
	unsigned int slab_size;
	char name[8];
};
static DEFINE_MUTEX(bio_slab_lock);
static struct bio_slab *bio_slabs;
static unsigned int bio_slab_nr, bio_slab_max;

static struct kmem_cache *bio_find_or_create_slab(unsigned int extra_size)
{
	unsigned int sz = sizeof(struct bio) + extra_size;
	struct kmem_cache *slab = NULL;
	struct bio_slab *bslab;
	unsigned int i, entry = -1;

	mutex_lock(&bio_slab_lock);

	i = 0;
	while (i < bio_slab_nr) {
		struct bio_slab *bslab = &bio_slabs[i];

		if (!bslab->slab && entry == -1)
			entry = i;
		else if (bslab->slab_size == sz) {
			slab = bslab->slab;
			bslab->slab_ref++;
			break;
		}
		i++;
	}

	if (slab)
		goto out_unlock;

	if (bio_slab_nr == bio_slab_max && entry == -1) {
		bio_slab_max <<= 1;
		bio_slabs = krealloc(bio_slabs,
				     bio_slab_max * sizeof(struct bio_slab),
				     GFP_KERNEL);
		if (!bio_slabs)
			goto out_unlock;
	}
	if (entry == -1)
		entry = bio_slab_nr++;

	bslab = &bio_slabs[entry];

	snprintf(bslab->name, sizeof(bslab->name), "bio-%d", entry);
	slab = kmem_cache_create(bslab->name, sz, 0, SLAB_HWCACHE_ALIGN, NULL);
	if (!slab)
		goto out_unlock;

	printk("bio: create slab <%s> at %d\n", bslab->name, entry);
	bslab->slab = slab;
	bslab->slab_ref = 1;
	bslab->slab_size = sz;
out_unlock:
	mutex_unlock(&bio_slab_lock);
	return slab;
}

static void bio_put_slab(struct bio_set *bs)
{
	struct bio_slab *bslab = NULL;
	unsigned int i;

	mutex_lock(&bio_slab_lock);

	for (i = 0; i < bio_slab_nr; i++) {
		if (bs->bio_slab == bio_slabs[i].slab) {
			bslab = &bio_slabs[i];
			break;
		}
	}

	if (WARN(!bslab, KERN_ERR "bio: unable to find slab!\n"))
		goto out;

	WARN_ON(!bslab->slab_ref);

	if (--bslab->slab_ref)
		goto out;

	kmem_cache_destroy(bslab->slab);
	bslab->slab = NULL;

out:
	mutex_unlock(&bio_slab_lock);
}

154 155 156 157 158
unsigned int bvec_nr_vecs(unsigned short idx)
{
	return bvec_slabs[idx].nr_vecs;
}

159 160 161 162 163 164 165 166 167 168 169 170 171
void bvec_free_bs(struct bio_set *bs, struct bio_vec *bv, unsigned int idx)
{
	BIO_BUG_ON(idx >= BIOVEC_NR_POOLS);

	if (idx == BIOVEC_MAX_IDX)
		mempool_free(bv, bs->bvec_pool);
	else {
		struct biovec_slab *bvs = bvec_slabs + idx;

		kmem_cache_free(bvs->slab, bv);
	}
}

172 173
struct bio_vec *bvec_alloc_bs(gfp_t gfp_mask, int nr, unsigned long *idx,
			      struct bio_set *bs)
L
Linus Torvalds 已提交
174 175 176 177
{
	struct bio_vec *bvl;

	/*
J
Jens Axboe 已提交
178 179 180
	 * If 'bs' is given, lookup the pool and do the mempool alloc.
	 * If not, this is a bio_kmalloc() allocation and just do a
	 * kzalloc() for the exact number of vecs right away.
L
Linus Torvalds 已提交
181
	 */
182
	if (!bs)
J
Jens Axboe 已提交
183
		bvl = kmalloc(nr * sizeof(struct bio_vec), gfp_mask);
184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221

	/*
	 * see comment near bvec_array define!
	 */
	switch (nr) {
	case 1:
		*idx = 0;
		break;
	case 2 ... 4:
		*idx = 1;
		break;
	case 5 ... 16:
		*idx = 2;
		break;
	case 17 ... 64:
		*idx = 3;
		break;
	case 65 ... 128:
		*idx = 4;
		break;
	case 129 ... BIO_MAX_PAGES:
		*idx = 5;
		break;
	default:
		return NULL;
	}

	/*
	 * idx now points to the pool we want to allocate from. only the
	 * 1-vec entry pool is mempool backed.
	 */
	if (*idx == BIOVEC_MAX_IDX) {
fallback:
		bvl = mempool_alloc(bs->bvec_pool, gfp_mask);
	} else {
		struct biovec_slab *bvs = bvec_slabs + *idx;
		gfp_t __gfp_mask = gfp_mask & ~(__GFP_WAIT | __GFP_IO);

J
Jens Axboe 已提交
222
		/*
223 224 225
		 * Make this allocation restricted and don't dump info on
		 * allocation failures, since we'll fallback to the mempool
		 * in case of failure.
J
Jens Axboe 已提交
226
		 */
227
		__gfp_mask |= __GFP_NOMEMALLOC | __GFP_NORETRY | __GFP_NOWARN;
L
Linus Torvalds 已提交
228

J
Jens Axboe 已提交
229
		/*
230 231
		 * Try a slab allocation. If this fails and __GFP_WAIT
		 * is set, retry with the 1-entry mempool
J
Jens Axboe 已提交
232
		 */
233 234 235 236 237 238 239
		bvl = kmem_cache_alloc(bvs->slab, __gfp_mask);
		if (unlikely(!bvl && (gfp_mask & __GFP_WAIT))) {
			*idx = BIOVEC_MAX_IDX;
			goto fallback;
		}
	}

L
Linus Torvalds 已提交
240 241 242
	return bvl;
}

243
void bio_free(struct bio *bio, struct bio_set *bs)
L
Linus Torvalds 已提交
244
{
245
	void *p;
L
Linus Torvalds 已提交
246

247
	if (bio_has_allocated_vec(bio))
248
		bvec_free_bs(bs, bio->bi_io_vec, BIO_POOL_IDX(bio));
L
Linus Torvalds 已提交
249

250
	if (bio_integrity(bio))
251
		bio_integrity_free(bio);
252

253 254 255 256 257 258 259 260
	/*
	 * If we have front padding, adjust the bio pointer before freeing
	 */
	p = bio;
	if (bs->front_pad)
		p -= bs->front_pad;

	mempool_free(p, bs->bio_pool);
P
Peter Osterlund 已提交
261 262 263 264 265 266 267 268
}

/*
 * default destructor for a bio allocated with bio_alloc_bioset()
 */
static void bio_fs_destructor(struct bio *bio)
{
	bio_free(bio, fs_bio_set);
L
Linus Torvalds 已提交
269 270
}

J
Jens Axboe 已提交
271 272
static void bio_kmalloc_destructor(struct bio *bio)
{
273 274
	if (bio_has_allocated_vec(bio))
		kfree(bio->bi_io_vec);
J
Jens Axboe 已提交
275 276 277
	kfree(bio);
}

278
void bio_init(struct bio *bio)
L
Linus Torvalds 已提交
279
{
J
Jens Axboe 已提交
280
	memset(bio, 0, sizeof(*bio));
L
Linus Torvalds 已提交
281
	bio->bi_flags = 1 << BIO_UPTODATE;
282
	bio->bi_comp_cpu = -1;
L
Linus Torvalds 已提交
283 284 285 286 287 288 289
	atomic_set(&bio->bi_cnt, 1);
}

/**
 * bio_alloc_bioset - allocate a bio for I/O
 * @gfp_mask:   the GFP_ mask given to the slab allocator
 * @nr_iovecs:	number of iovecs to pre-allocate
J
Jens Axboe 已提交
290
 * @bs:		the bio_set to allocate from. If %NULL, just use kmalloc
L
Linus Torvalds 已提交
291 292
 *
 * Description:
J
Jens Axboe 已提交
293
 *   bio_alloc_bioset will first try its own mempool to satisfy the allocation.
L
Linus Torvalds 已提交
294
 *   If %__GFP_WAIT is set then we will block on the internal pool waiting
J
Jens Axboe 已提交
295 296
 *   for a &struct bio to become free. If a %NULL @bs is passed in, we will
 *   fall back to just using @kmalloc to allocate the required memory.
L
Linus Torvalds 已提交
297
 *
298 299 300
 *   Note that the caller must set ->bi_destructor on succesful return
 *   of a bio, to do the appropriate freeing of the bio once the reference
 *   count drops to zero.
L
Linus Torvalds 已提交
301
 **/
A
Al Viro 已提交
302
struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
L
Linus Torvalds 已提交
303
{
I
Ingo Molnar 已提交
304
	struct bio_vec *bvl = NULL;
305
	struct bio *bio = NULL;
I
Ingo Molnar 已提交
306 307
	unsigned long idx = 0;
	void *p = NULL;
308 309

	if (bs) {
310
		p = mempool_alloc(bs->bio_pool, gfp_mask);
I
Ingo Molnar 已提交
311 312 313 314
		if (!p)
			goto err;
		bio = p + bs->front_pad;
	} else {
J
Jens Axboe 已提交
315
		bio = kmalloc(sizeof(*bio), gfp_mask);
I
Ingo Molnar 已提交
316 317 318
		if (!bio)
			goto err;
	}
L
Linus Torvalds 已提交
319

I
Ingo Molnar 已提交
320 321 322 323 324 325 326 327 328 329 330 331 332 333
	bio_init(bio);

	if (unlikely(!nr_iovecs))
		goto out_set;

	if (nr_iovecs <= BIO_INLINE_VECS) {
		bvl = bio->bi_inline_vecs;
		nr_iovecs = BIO_INLINE_VECS;
	} else {
		bvl = bvec_alloc_bs(gfp_mask, nr_iovecs, &idx, bs);
		if (unlikely(!bvl))
			goto err_free;

		nr_iovecs = bvec_nr_vecs(idx);
L
Linus Torvalds 已提交
334
	}
I
Ingo Molnar 已提交
335 336 337 338 339
	bio->bi_flags |= idx << BIO_POOL_OFFSET;
	bio->bi_max_vecs = nr_iovecs;
out_set:
	bio->bi_io_vec = bvl;

L
Linus Torvalds 已提交
340
	return bio;
I
Ingo Molnar 已提交
341 342 343 344 345 346 347 348

err_free:
	if (bs)
		mempool_free(p, bs->bio_pool);
	else
		kfree(bio);
err:
	return NULL;
L
Linus Torvalds 已提交
349 350
}

351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368
/**
 * bio_alloc - allocate a bio for I/O
 * @gfp_mask:   the GFP_ mask given to the slab allocator
 * @nr_iovecs:	number of iovecs to pre-allocate
 *
 * Description:
 *   bio_alloc will allocate a bio and associated bio_vec array that can hold
 *   at least @nr_iovecs entries. Allocations will be done from the
 *   fs_bio_set. Also see @bio_alloc_bioset.
 *
 *   If %__GFP_WAIT is set, then bio_alloc will always be able to allocate
 *   a bio. This is due to the mempool guarantees. To make this work, callers
 *   must never allocate more than 1 bio at the time from this pool. Callers
 *   that need to allocate more than 1 bio must always submit the previously
 *   allocate bio for IO before attempting to allocate a new one. Failure to
 *   do so can cause livelocks under memory pressure.
 *
 **/
A
Al Viro 已提交
369
struct bio *bio_alloc(gfp_t gfp_mask, int nr_iovecs)
L
Linus Torvalds 已提交
370
{
P
Peter Osterlund 已提交
371 372 373 374 375 376
	struct bio *bio = bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);

	if (bio)
		bio->bi_destructor = bio_fs_destructor;

	return bio;
L
Linus Torvalds 已提交
377 378
}

J
Jens Axboe 已提交
379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395
/*
 * Like bio_alloc(), but doesn't use a mempool backing. This means that
 * it CAN fail, but while bio_alloc() can only be used for allocations
 * that have a short (finite) life span, bio_kmalloc() should be used
 * for more permanent bio allocations (like allocating some bio's for
 * initalization or setup purposes).
 */
struct bio *bio_kmalloc(gfp_t gfp_mask, int nr_iovecs)
{
	struct bio *bio = bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);

	if (bio)
		bio->bi_destructor = bio_kmalloc_destructor;

	return bio;
}

L
Linus Torvalds 已提交
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431
void zero_fill_bio(struct bio *bio)
{
	unsigned long flags;
	struct bio_vec *bv;
	int i;

	bio_for_each_segment(bv, bio, i) {
		char *data = bvec_kmap_irq(bv, &flags);
		memset(data, 0, bv->bv_len);
		flush_dcache_page(bv->bv_page);
		bvec_kunmap_irq(data, &flags);
	}
}
EXPORT_SYMBOL(zero_fill_bio);

/**
 * bio_put - release a reference to a bio
 * @bio:   bio to release reference to
 *
 * Description:
 *   Put a reference to a &struct bio, either one you have gotten with
 *   bio_alloc or bio_get. The last put of a bio will free it.
 **/
void bio_put(struct bio *bio)
{
	BIO_BUG_ON(!atomic_read(&bio->bi_cnt));

	/*
	 * last put frees it
	 */
	if (atomic_dec_and_test(&bio->bi_cnt)) {
		bio->bi_next = NULL;
		bio->bi_destructor(bio);
	}
}

432
inline int bio_phys_segments(struct request_queue *q, struct bio *bio)
L
Linus Torvalds 已提交
433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448
{
	if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
		blk_recount_segments(q, bio);

	return bio->bi_phys_segments;
}

/**
 * 	__bio_clone	-	clone a bio
 * 	@bio: destination bio
 * 	@bio_src: bio to clone
 *
 *	Clone a &bio. Caller will own the returned bio, but not
 *	the actual data it points to. Reference count of returned
 * 	bio will be one.
 */
449
void __bio_clone(struct bio *bio, struct bio *bio_src)
L
Linus Torvalds 已提交
450
{
451 452
	memcpy(bio->bi_io_vec, bio_src->bi_io_vec,
		bio_src->bi_max_vecs * sizeof(struct bio_vec));
L
Linus Torvalds 已提交
453

454 455 456 457
	/*
	 * most users will be overriding ->bi_bdev with a new target,
	 * so we don't set nor calculate new physical/hw segment counts here
	 */
L
Linus Torvalds 已提交
458 459 460 461 462 463
	bio->bi_sector = bio_src->bi_sector;
	bio->bi_bdev = bio_src->bi_bdev;
	bio->bi_flags |= 1 << BIO_CLONED;
	bio->bi_rw = bio_src->bi_rw;
	bio->bi_vcnt = bio_src->bi_vcnt;
	bio->bi_size = bio_src->bi_size;
A
Andrew Morton 已提交
464
	bio->bi_idx = bio_src->bi_idx;
L
Linus Torvalds 已提交
465 466 467 468 469 470 471 472 473
}

/**
 *	bio_clone	-	clone a bio
 *	@bio: bio to clone
 *	@gfp_mask: allocation priority
 *
 * 	Like __bio_clone, only also allocates the returned bio
 */
A
Al Viro 已提交
474
struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
L
Linus Torvalds 已提交
475 476 477
{
	struct bio *b = bio_alloc_bioset(gfp_mask, bio->bi_max_vecs, fs_bio_set);

478 479 480 481 482 483 484 485 486
	if (!b)
		return NULL;

	b->bi_destructor = bio_fs_destructor;
	__bio_clone(b, bio);

	if (bio_integrity(bio)) {
		int ret;

487
		ret = bio_integrity_clone(b, bio, gfp_mask);
488

L
Li Zefan 已提交
489 490
		if (ret < 0) {
			bio_put(b);
491
			return NULL;
L
Li Zefan 已提交
492
		}
P
Peter Osterlund 已提交
493
	}
L
Linus Torvalds 已提交
494 495 496 497 498 499 500 501 502 503 504 505 506 507 508

	return b;
}

/**
 *	bio_get_nr_vecs		- return approx number of vecs
 *	@bdev:  I/O target
 *
 *	Return the approximate number of pages we can send to this target.
 *	There's no guarantee that you will be able to fit this number of pages
 *	into a bio, it does not account for dynamic restrictions that vary
 *	on offset.
 */
int bio_get_nr_vecs(struct block_device *bdev)
{
509
	struct request_queue *q = bdev_get_queue(bdev);
L
Linus Torvalds 已提交
510 511 512 513 514 515 516 517 518 519 520
	int nr_pages;

	nr_pages = ((q->max_sectors << 9) + PAGE_SIZE - 1) >> PAGE_SHIFT;
	if (nr_pages > q->max_phys_segments)
		nr_pages = q->max_phys_segments;
	if (nr_pages > q->max_hw_segments)
		nr_pages = q->max_hw_segments;

	return nr_pages;
}

521
static int __bio_add_page(struct request_queue *q, struct bio *bio, struct page
522 523
			  *page, unsigned int len, unsigned int offset,
			  unsigned short max_sectors)
L
Linus Torvalds 已提交
524 525 526 527 528 529 530 531 532 533
{
	int retried_segments = 0;
	struct bio_vec *bvec;

	/*
	 * cloned bio must not modify vec list
	 */
	if (unlikely(bio_flagged(bio, BIO_CLONED)))
		return 0;

534
	if (((bio->bi_size + len) >> 9) > max_sectors)
L
Linus Torvalds 已提交
535 536
		return 0;

537 538 539 540 541 542 543 544 545 546 547
	/*
	 * For filesystems with a blocksize smaller than the pagesize
	 * we will often be called with the same page as last time and
	 * a consecutive offset.  Optimize this special case.
	 */
	if (bio->bi_vcnt > 0) {
		struct bio_vec *prev = &bio->bi_io_vec[bio->bi_vcnt - 1];

		if (page == prev->bv_page &&
		    offset == prev->bv_offset + prev->bv_len) {
			prev->bv_len += len;
548 549 550 551 552 553 554 555 556 557 558 559 560

			if (q->merge_bvec_fn) {
				struct bvec_merge_data bvm = {
					.bi_bdev = bio->bi_bdev,
					.bi_sector = bio->bi_sector,
					.bi_size = bio->bi_size,
					.bi_rw = bio->bi_rw,
				};

				if (q->merge_bvec_fn(q, &bvm, prev) < len) {
					prev->bv_len -= len;
					return 0;
				}
561 562 563 564 565 566 567
			}

			goto done;
		}
	}

	if (bio->bi_vcnt >= bio->bi_max_vecs)
L
Linus Torvalds 已提交
568 569 570 571 572 573 574 575
		return 0;

	/*
	 * we might lose a segment or two here, but rather that than
	 * make this too complex.
	 */

	while (bio->bi_phys_segments >= q->max_phys_segments
M
Mikulas Patocka 已提交
576
	       || bio->bi_phys_segments >= q->max_hw_segments) {
L
Linus Torvalds 已提交
577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599

		if (retried_segments)
			return 0;

		retried_segments = 1;
		blk_recount_segments(q, bio);
	}

	/*
	 * setup the new entry, we might clear it again later if we
	 * cannot add the page
	 */
	bvec = &bio->bi_io_vec[bio->bi_vcnt];
	bvec->bv_page = page;
	bvec->bv_len = len;
	bvec->bv_offset = offset;

	/*
	 * if queue has other restrictions (eg varying max sector size
	 * depending on offset), it can specify a merge_bvec_fn in the
	 * queue to get further control
	 */
	if (q->merge_bvec_fn) {
600 601 602 603 604 605 606
		struct bvec_merge_data bvm = {
			.bi_bdev = bio->bi_bdev,
			.bi_sector = bio->bi_sector,
			.bi_size = bio->bi_size,
			.bi_rw = bio->bi_rw,
		};

L
Linus Torvalds 已提交
607 608 609 610
		/*
		 * merge_bvec_fn() returns number of bytes it can accept
		 * at this offset
		 */
611
		if (q->merge_bvec_fn(q, &bvm, bvec) < len) {
L
Linus Torvalds 已提交
612 613 614 615 616 617 618 619
			bvec->bv_page = NULL;
			bvec->bv_len = 0;
			bvec->bv_offset = 0;
			return 0;
		}
	}

	/* If we may be able to merge these biovecs, force a recount */
620
	if (bio->bi_vcnt && (BIOVEC_PHYS_MERGEABLE(bvec-1, bvec)))
L
Linus Torvalds 已提交
621 622 623 624
		bio->bi_flags &= ~(1 << BIO_SEG_VALID);

	bio->bi_vcnt++;
	bio->bi_phys_segments++;
625
 done:
L
Linus Torvalds 已提交
626 627 628 629
	bio->bi_size += len;
	return len;
}

630 631
/**
 *	bio_add_pc_page	-	attempt to add page to bio
J
Jens Axboe 已提交
632
 *	@q: the target queue
633 634 635 636 637 638 639 640 641 642 643
 *	@bio: destination bio
 *	@page: page to add
 *	@len: vec entry length
 *	@offset: vec entry offset
 *
 *	Attempt to add a page to the bio_vec maplist. This can fail for a
 *	number of reasons, such as the bio being full or target block
 *	device limitations. The target block device must allow bio's
 *      smaller than PAGE_SIZE, so it is always possible to add a single
 *      page to an empty bio. This should only be used by REQ_PC bios.
 */
644
int bio_add_pc_page(struct request_queue *q, struct bio *bio, struct page *page,
645 646
		    unsigned int len, unsigned int offset)
{
647
	return __bio_add_page(q, bio, page, len, offset, q->max_hw_sectors);
648 649
}

L
Linus Torvalds 已提交
650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665
/**
 *	bio_add_page	-	attempt to add page to bio
 *	@bio: destination bio
 *	@page: page to add
 *	@len: vec entry length
 *	@offset: vec entry offset
 *
 *	Attempt to add a page to the bio_vec maplist. This can fail for a
 *	number of reasons, such as the bio being full or target block
 *	device limitations. The target block device must allow bio's
 *      smaller than PAGE_SIZE, so it is always possible to add a single
 *      page to an empty bio.
 */
int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
		 unsigned int offset)
{
666 667
	struct request_queue *q = bdev_get_queue(bio->bi_bdev);
	return __bio_add_page(q, bio, page, len, offset, q->max_sectors);
L
Linus Torvalds 已提交
668 669 670 671
}

struct bio_map_data {
	struct bio_vec *iovecs;
672
	struct sg_iovec *sgvecs;
673 674
	int nr_sgvecs;
	int is_our_pages;
L
Linus Torvalds 已提交
675 676
};

677
static void bio_set_map_data(struct bio_map_data *bmd, struct bio *bio,
678 679
			     struct sg_iovec *iov, int iov_count,
			     int is_our_pages)
L
Linus Torvalds 已提交
680 681
{
	memcpy(bmd->iovecs, bio->bi_io_vec, sizeof(struct bio_vec) * bio->bi_vcnt);
682 683
	memcpy(bmd->sgvecs, iov, sizeof(struct sg_iovec) * iov_count);
	bmd->nr_sgvecs = iov_count;
684
	bmd->is_our_pages = is_our_pages;
L
Linus Torvalds 已提交
685 686 687 688 689 690
	bio->bi_private = bmd;
}

static void bio_free_map_data(struct bio_map_data *bmd)
{
	kfree(bmd->iovecs);
691
	kfree(bmd->sgvecs);
L
Linus Torvalds 已提交
692 693 694
	kfree(bmd);
}

695 696
static struct bio_map_data *bio_alloc_map_data(int nr_segs, int iov_count,
					       gfp_t gfp_mask)
L
Linus Torvalds 已提交
697
{
698
	struct bio_map_data *bmd = kmalloc(sizeof(*bmd), gfp_mask);
L
Linus Torvalds 已提交
699 700 701 702

	if (!bmd)
		return NULL;

703
	bmd->iovecs = kmalloc(sizeof(struct bio_vec) * nr_segs, gfp_mask);
704 705 706 707 708
	if (!bmd->iovecs) {
		kfree(bmd);
		return NULL;
	}

709
	bmd->sgvecs = kmalloc(sizeof(struct sg_iovec) * iov_count, gfp_mask);
710
	if (bmd->sgvecs)
L
Linus Torvalds 已提交
711 712
		return bmd;

713
	kfree(bmd->iovecs);
L
Linus Torvalds 已提交
714 715 716 717
	kfree(bmd);
	return NULL;
}

718
static int __bio_copy_iov(struct bio *bio, struct bio_vec *iovecs,
719 720
			  struct sg_iovec *iov, int iov_count, int uncopy,
			  int do_free_page)
721 722 723 724 725 726 727 728 729
{
	int ret = 0, i;
	struct bio_vec *bvec;
	int iov_idx = 0;
	unsigned int iov_off = 0;
	int read = bio_data_dir(bio) == READ;

	__bio_for_each_segment(bvec, bio, i, 0) {
		char *bv_addr = page_address(bvec->bv_page);
730
		unsigned int bv_len = iovecs[i].bv_len;
731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762

		while (bv_len && iov_idx < iov_count) {
			unsigned int bytes;
			char *iov_addr;

			bytes = min_t(unsigned int,
				      iov[iov_idx].iov_len - iov_off, bv_len);
			iov_addr = iov[iov_idx].iov_base + iov_off;

			if (!ret) {
				if (!read && !uncopy)
					ret = copy_from_user(bv_addr, iov_addr,
							     bytes);
				if (read && uncopy)
					ret = copy_to_user(iov_addr, bv_addr,
							   bytes);

				if (ret)
					ret = -EFAULT;
			}

			bv_len -= bytes;
			bv_addr += bytes;
			iov_addr += bytes;
			iov_off += bytes;

			if (iov[iov_idx].iov_len == iov_off) {
				iov_idx++;
				iov_off = 0;
			}
		}

763
		if (do_free_page)
764 765 766 767 768 769
			__free_page(bvec->bv_page);
	}

	return ret;
}

L
Linus Torvalds 已提交
770 771 772 773 774 775 776 777 778 779
/**
 *	bio_uncopy_user	-	finish previously mapped bio
 *	@bio: bio being terminated
 *
 *	Free pages allocated from bio_copy_user() and write back data
 *	to user space in case of a read.
 */
int bio_uncopy_user(struct bio *bio)
{
	struct bio_map_data *bmd = bio->bi_private;
780
	int ret = 0;
L
Linus Torvalds 已提交
781

782 783 784
	if (!bio_flagged(bio, BIO_NULL_MAPPED))
		ret = __bio_copy_iov(bio, bmd->iovecs, bmd->sgvecs,
				     bmd->nr_sgvecs, 1, bmd->is_our_pages);
L
Linus Torvalds 已提交
785 786 787 788 789 790
	bio_free_map_data(bmd);
	bio_put(bio);
	return ret;
}

/**
791
 *	bio_copy_user_iov	-	copy user data to bio
L
Linus Torvalds 已提交
792
 *	@q: destination block queue
793
 *	@map_data: pointer to the rq_map_data holding pages (if necessary)
794 795
 *	@iov:	the iovec.
 *	@iov_count: number of elements in the iovec
L
Linus Torvalds 已提交
796
 *	@write_to_vm: bool indicating writing to pages or not
797
 *	@gfp_mask: memory allocation flags
L
Linus Torvalds 已提交
798 799 800 801 802
 *
 *	Prepares and returns a bio for indirect user io, bouncing data
 *	to/from kernel pages as necessary. Must be paired with
 *	call bio_uncopy_user() on io completion.
 */
803 804 805 806
struct bio *bio_copy_user_iov(struct request_queue *q,
			      struct rq_map_data *map_data,
			      struct sg_iovec *iov, int iov_count,
			      int write_to_vm, gfp_t gfp_mask)
L
Linus Torvalds 已提交
807 808 809 810 811 812
{
	struct bio_map_data *bmd;
	struct bio_vec *bvec;
	struct page *page;
	struct bio *bio;
	int i, ret;
813 814
	int nr_pages = 0;
	unsigned int len = 0;
815
	unsigned int offset = map_data ? map_data->offset & ~PAGE_MASK : 0;
L
Linus Torvalds 已提交
816

817 818 819 820 821 822 823 824 825 826 827 828 829
	for (i = 0; i < iov_count; i++) {
		unsigned long uaddr;
		unsigned long end;
		unsigned long start;

		uaddr = (unsigned long)iov[i].iov_base;
		end = (uaddr + iov[i].iov_len + PAGE_SIZE - 1) >> PAGE_SHIFT;
		start = uaddr >> PAGE_SHIFT;

		nr_pages += end - start;
		len += iov[i].iov_len;
	}

830
	bmd = bio_alloc_map_data(nr_pages, iov_count, gfp_mask);
L
Linus Torvalds 已提交
831 832 833 834
	if (!bmd)
		return ERR_PTR(-ENOMEM);

	ret = -ENOMEM;
835
	bio = bio_alloc(gfp_mask, nr_pages);
L
Linus Torvalds 已提交
836 837 838 839 840 841
	if (!bio)
		goto out_bmd;

	bio->bi_rw |= (!write_to_vm << BIO_RW);

	ret = 0;
842 843

	if (map_data) {
844
		nr_pages = 1 << map_data->page_order;
845 846
		i = map_data->offset / PAGE_SIZE;
	}
L
Linus Torvalds 已提交
847
	while (len) {
848
		unsigned int bytes = PAGE_SIZE;
L
Linus Torvalds 已提交
849

850 851
		bytes -= offset;

L
Linus Torvalds 已提交
852 853 854
		if (bytes > len)
			bytes = len;

855
		if (map_data) {
856
			if (i == map_data->nr_entries * nr_pages) {
857 858 859
				ret = -ENOMEM;
				break;
			}
860 861 862 863 864 865

			page = map_data->pages[i / nr_pages];
			page += (i % nr_pages);

			i++;
		} else {
866
			page = alloc_page(q->bounce_gfp | gfp_mask);
867 868 869 870
			if (!page) {
				ret = -ENOMEM;
				break;
			}
L
Linus Torvalds 已提交
871 872
		}

873
		if (bio_add_pc_page(q, bio, page, bytes, offset) < bytes)
L
Linus Torvalds 已提交
874 875 876
			break;

		len -= bytes;
877
		offset = 0;
L
Linus Torvalds 已提交
878 879 880 881 882 883 884 885
	}

	if (ret)
		goto cleanup;

	/*
	 * success
	 */
886
	if (!write_to_vm && (!map_data || !map_data->null_mapped)) {
887
		ret = __bio_copy_iov(bio, bio->bi_io_vec, iov, iov_count, 0, 0);
888 889
		if (ret)
			goto cleanup;
L
Linus Torvalds 已提交
890 891
	}

892
	bio_set_map_data(bmd, bio, iov, iov_count, map_data ? 0 : 1);
L
Linus Torvalds 已提交
893 894
	return bio;
cleanup:
895 896 897
	if (!map_data)
		bio_for_each_segment(bvec, bio, i)
			__free_page(bvec->bv_page);
L
Linus Torvalds 已提交
898 899 900 901 902 903 904

	bio_put(bio);
out_bmd:
	bio_free_map_data(bmd);
	return ERR_PTR(ret);
}

905 906 907
/**
 *	bio_copy_user	-	copy user data to bio
 *	@q: destination block queue
908
 *	@map_data: pointer to the rq_map_data holding pages (if necessary)
909 910 911
 *	@uaddr: start of user address
 *	@len: length in bytes
 *	@write_to_vm: bool indicating writing to pages or not
912
 *	@gfp_mask: memory allocation flags
913 914 915 916 917
 *
 *	Prepares and returns a bio for indirect user io, bouncing data
 *	to/from kernel pages as necessary. Must be paired with
 *	call bio_uncopy_user() on io completion.
 */
918 919 920
struct bio *bio_copy_user(struct request_queue *q, struct rq_map_data *map_data,
			  unsigned long uaddr, unsigned int len,
			  int write_to_vm, gfp_t gfp_mask)
921 922 923 924 925 926
{
	struct sg_iovec iov;

	iov.iov_base = (void __user *)uaddr;
	iov.iov_len = len;

927
	return bio_copy_user_iov(q, map_data, &iov, 1, write_to_vm, gfp_mask);
928 929
}

930
static struct bio *__bio_map_user_iov(struct request_queue *q,
931 932
				      struct block_device *bdev,
				      struct sg_iovec *iov, int iov_count,
933
				      int write_to_vm, gfp_t gfp_mask)
L
Linus Torvalds 已提交
934
{
935 936
	int i, j;
	int nr_pages = 0;
L
Linus Torvalds 已提交
937 938
	struct page **pages;
	struct bio *bio;
939 940
	int cur_page = 0;
	int ret, offset;
L
Linus Torvalds 已提交
941

942 943 944 945 946 947 948 949
	for (i = 0; i < iov_count; i++) {
		unsigned long uaddr = (unsigned long)iov[i].iov_base;
		unsigned long len = iov[i].iov_len;
		unsigned long end = (uaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
		unsigned long start = uaddr >> PAGE_SHIFT;

		nr_pages += end - start;
		/*
950
		 * buffer must be aligned to at least hardsector size for now
951
		 */
952
		if (uaddr & queue_dma_alignment(q))
953 954 955 956
			return ERR_PTR(-EINVAL);
	}

	if (!nr_pages)
L
Linus Torvalds 已提交
957 958
		return ERR_PTR(-EINVAL);

959
	bio = bio_alloc(gfp_mask, nr_pages);
L
Linus Torvalds 已提交
960 961 962 963
	if (!bio)
		return ERR_PTR(-ENOMEM);

	ret = -ENOMEM;
964
	pages = kcalloc(nr_pages, sizeof(struct page *), gfp_mask);
L
Linus Torvalds 已提交
965 966 967
	if (!pages)
		goto out;

968 969 970 971 972 973 974 975
	for (i = 0; i < iov_count; i++) {
		unsigned long uaddr = (unsigned long)iov[i].iov_base;
		unsigned long len = iov[i].iov_len;
		unsigned long end = (uaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
		unsigned long start = uaddr >> PAGE_SHIFT;
		const int local_nr_pages = end - start;
		const int page_limit = cur_page + local_nr_pages;
		
N
Nick Piggin 已提交
976 977
		ret = get_user_pages_fast(uaddr, local_nr_pages,
				write_to_vm, &pages[cur_page]);
978 979
		if (ret < local_nr_pages) {
			ret = -EFAULT;
980
			goto out_unmap;
981
		}
982 983 984 985 986 987 988 989 990 991 992 993 994 995

		offset = uaddr & ~PAGE_MASK;
		for (j = cur_page; j < page_limit; j++) {
			unsigned int bytes = PAGE_SIZE - offset;

			if (len <= 0)
				break;
			
			if (bytes > len)
				bytes = len;

			/*
			 * sorry...
			 */
996 997
			if (bio_add_pc_page(q, bio, pages[j], bytes, offset) <
					    bytes)
998 999 1000 1001 1002
				break;

			len -= bytes;
			offset = 0;
		}
L
Linus Torvalds 已提交
1003

1004
		cur_page = j;
L
Linus Torvalds 已提交
1005
		/*
1006
		 * release the pages we didn't map into the bio, if any
L
Linus Torvalds 已提交
1007
		 */
1008 1009
		while (j < page_limit)
			page_cache_release(pages[j++]);
L
Linus Torvalds 已提交
1010 1011 1012 1013 1014 1015 1016 1017 1018 1019
	}

	kfree(pages);

	/*
	 * set data direction, and check if mapped pages need bouncing
	 */
	if (!write_to_vm)
		bio->bi_rw |= (1 << BIO_RW);

1020
	bio->bi_bdev = bdev;
L
Linus Torvalds 已提交
1021 1022
	bio->bi_flags |= (1 << BIO_USER_MAPPED);
	return bio;
1023 1024 1025 1026 1027 1028 1029 1030

 out_unmap:
	for (i = 0; i < nr_pages; i++) {
		if(!pages[i])
			break;
		page_cache_release(pages[i]);
	}
 out:
L
Linus Torvalds 已提交
1031 1032 1033 1034 1035 1036 1037
	kfree(pages);
	bio_put(bio);
	return ERR_PTR(ret);
}

/**
 *	bio_map_user	-	map user address into bio
1038
 *	@q: the struct request_queue for the bio
L
Linus Torvalds 已提交
1039 1040 1041 1042
 *	@bdev: destination block device
 *	@uaddr: start of user address
 *	@len: length in bytes
 *	@write_to_vm: bool indicating writing to pages or not
1043
 *	@gfp_mask: memory allocation flags
L
Linus Torvalds 已提交
1044 1045 1046 1047
 *
 *	Map the user space address into a bio suitable for io to a block
 *	device. Returns an error pointer in case of error.
 */
1048
struct bio *bio_map_user(struct request_queue *q, struct block_device *bdev,
1049 1050
			 unsigned long uaddr, unsigned int len, int write_to_vm,
			 gfp_t gfp_mask)
1051 1052 1053
{
	struct sg_iovec iov;

1054
	iov.iov_base = (void __user *)uaddr;
1055 1056
	iov.iov_len = len;

1057
	return bio_map_user_iov(q, bdev, &iov, 1, write_to_vm, gfp_mask);
1058 1059 1060 1061
}

/**
 *	bio_map_user_iov - map user sg_iovec table into bio
1062
 *	@q: the struct request_queue for the bio
1063 1064 1065 1066
 *	@bdev: destination block device
 *	@iov:	the iovec.
 *	@iov_count: number of elements in the iovec
 *	@write_to_vm: bool indicating writing to pages or not
1067
 *	@gfp_mask: memory allocation flags
1068 1069 1070 1071
 *
 *	Map the user space address into a bio suitable for io to a block
 *	device. Returns an error pointer in case of error.
 */
1072
struct bio *bio_map_user_iov(struct request_queue *q, struct block_device *bdev,
1073
			     struct sg_iovec *iov, int iov_count,
1074
			     int write_to_vm, gfp_t gfp_mask)
L
Linus Torvalds 已提交
1075 1076 1077
{
	struct bio *bio;

1078 1079
	bio = __bio_map_user_iov(q, bdev, iov, iov_count, write_to_vm,
				 gfp_mask);
L
Linus Torvalds 已提交
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
	if (IS_ERR(bio))
		return bio;

	/*
	 * subtle -- if __bio_map_user() ended up bouncing a bio,
	 * it would normally disappear when its bi_end_io is run.
	 * however, we need it for the unmap, so grab an extra
	 * reference to it
	 */
	bio_get(bio);

1091
	return bio;
L
Linus Torvalds 已提交
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126
}

static void __bio_unmap_user(struct bio *bio)
{
	struct bio_vec *bvec;
	int i;

	/*
	 * make sure we dirty pages we wrote to
	 */
	__bio_for_each_segment(bvec, bio, i, 0) {
		if (bio_data_dir(bio) == READ)
			set_page_dirty_lock(bvec->bv_page);

		page_cache_release(bvec->bv_page);
	}

	bio_put(bio);
}

/**
 *	bio_unmap_user	-	unmap a bio
 *	@bio:		the bio being unmapped
 *
 *	Unmap a bio previously mapped by bio_map_user(). Must be called with
 *	a process context.
 *
 *	bio_unmap_user() may sleep.
 */
void bio_unmap_user(struct bio *bio)
{
	__bio_unmap_user(bio);
	bio_put(bio);
}

1127
static void bio_map_kern_endio(struct bio *bio, int err)
1128 1129 1130 1131 1132
{
	bio_put(bio);
}


1133
static struct bio *__bio_map_kern(struct request_queue *q, void *data,
A
Al Viro 已提交
1134
				  unsigned int len, gfp_t gfp_mask)
M
Mike Christie 已提交
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156
{
	unsigned long kaddr = (unsigned long)data;
	unsigned long end = (kaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
	unsigned long start = kaddr >> PAGE_SHIFT;
	const int nr_pages = end - start;
	int offset, i;
	struct bio *bio;

	bio = bio_alloc(gfp_mask, nr_pages);
	if (!bio)
		return ERR_PTR(-ENOMEM);

	offset = offset_in_page(kaddr);
	for (i = 0; i < nr_pages; i++) {
		unsigned int bytes = PAGE_SIZE - offset;

		if (len <= 0)
			break;

		if (bytes > len)
			bytes = len;

1157 1158
		if (bio_add_pc_page(q, bio, virt_to_page(data), bytes,
				    offset) < bytes)
M
Mike Christie 已提交
1159 1160 1161 1162 1163 1164 1165
			break;

		data += bytes;
		len -= bytes;
		offset = 0;
	}

1166
	bio->bi_end_io = bio_map_kern_endio;
M
Mike Christie 已提交
1167 1168 1169 1170 1171
	return bio;
}

/**
 *	bio_map_kern	-	map kernel address into bio
1172
 *	@q: the struct request_queue for the bio
M
Mike Christie 已提交
1173 1174 1175 1176 1177 1178 1179
 *	@data: pointer to buffer to map
 *	@len: length in bytes
 *	@gfp_mask: allocation flags for bio allocation
 *
 *	Map the kernel address into a bio suitable for io to a block
 *	device. Returns an error pointer in case of error.
 */
1180
struct bio *bio_map_kern(struct request_queue *q, void *data, unsigned int len,
A
Al Viro 已提交
1181
			 gfp_t gfp_mask)
M
Mike Christie 已提交
1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
{
	struct bio *bio;

	bio = __bio_map_kern(q, data, len, gfp_mask);
	if (IS_ERR(bio))
		return bio;

	if (bio->bi_size == len)
		return bio;

	/*
	 * Don't support partial mappings.
	 */
	bio_put(bio);
	return ERR_PTR(-EINVAL);
}

1199 1200 1201 1202
static void bio_copy_kern_endio(struct bio *bio, int err)
{
	struct bio_vec *bvec;
	const int read = bio_data_dir(bio) == READ;
1203
	struct bio_map_data *bmd = bio->bi_private;
1204
	int i;
1205
	char *p = bmd->sgvecs[0].iov_base;
1206 1207 1208

	__bio_for_each_segment(bvec, bio, i, 0) {
		char *addr = page_address(bvec->bv_page);
1209
		int len = bmd->iovecs[i].bv_len;
1210 1211

		if (read && !err)
1212
			memcpy(p, addr, len);
1213 1214

		__free_page(bvec->bv_page);
1215
		p += len;
1216 1217
	}

1218
	bio_free_map_data(bmd);
1219 1220 1221 1222 1223 1224 1225 1226 1227
	bio_put(bio);
}

/**
 *	bio_copy_kern	-	copy kernel address into bio
 *	@q: the struct request_queue for the bio
 *	@data: pointer to buffer to copy
 *	@len: length in bytes
 *	@gfp_mask: allocation flags for bio and page allocation
1228
 *	@reading: data direction is READ
1229 1230 1231 1232 1233 1234 1235 1236 1237
 *
 *	copy the kernel address into a bio suitable for io to a block
 *	device. Returns an error pointer in case of error.
 */
struct bio *bio_copy_kern(struct request_queue *q, void *data, unsigned int len,
			  gfp_t gfp_mask, int reading)
{
	struct bio *bio;
	struct bio_vec *bvec;
1238
	int i;
1239

1240 1241 1242
	bio = bio_copy_user(q, NULL, (unsigned long)data, len, 1, gfp_mask);
	if (IS_ERR(bio))
		return bio;
1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255

	if (!reading) {
		void *p = data;

		bio_for_each_segment(bvec, bio, i) {
			char *addr = page_address(bvec->bv_page);

			memcpy(addr, p, bvec->bv_len);
			p += bvec->bv_len;
		}
	}

	bio->bi_end_io = bio_copy_kern_endio;
1256

1257 1258 1259
	return bio;
}

L
Linus Torvalds 已提交
1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301
/*
 * bio_set_pages_dirty() and bio_check_pages_dirty() are support functions
 * for performing direct-IO in BIOs.
 *
 * The problem is that we cannot run set_page_dirty() from interrupt context
 * because the required locks are not interrupt-safe.  So what we can do is to
 * mark the pages dirty _before_ performing IO.  And in interrupt context,
 * check that the pages are still dirty.   If so, fine.  If not, redirty them
 * in process context.
 *
 * We special-case compound pages here: normally this means reads into hugetlb
 * pages.  The logic in here doesn't really work right for compound pages
 * because the VM does not uniformly chase down the head page in all cases.
 * But dirtiness of compound pages is pretty meaningless anyway: the VM doesn't
 * handle them at all.  So we skip compound pages here at an early stage.
 *
 * Note that this code is very hard to test under normal circumstances because
 * direct-io pins the pages with get_user_pages().  This makes
 * is_page_cache_freeable return false, and the VM will not clean the pages.
 * But other code (eg, pdflush) could clean the pages if they are mapped
 * pagecache.
 *
 * Simply disabling the call to bio_set_pages_dirty() is a good way to test the
 * deferred bio dirtying paths.
 */

/*
 * bio_set_pages_dirty() will mark all the bio's pages as dirty.
 */
void bio_set_pages_dirty(struct bio *bio)
{
	struct bio_vec *bvec = bio->bi_io_vec;
	int i;

	for (i = 0; i < bio->bi_vcnt; i++) {
		struct page *page = bvec[i].bv_page;

		if (page && !PageCompound(page))
			set_page_dirty_lock(page);
	}
}

1302
static void bio_release_pages(struct bio *bio)
L
Linus Torvalds 已提交
1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325
{
	struct bio_vec *bvec = bio->bi_io_vec;
	int i;

	for (i = 0; i < bio->bi_vcnt; i++) {
		struct page *page = bvec[i].bv_page;

		if (page)
			put_page(page);
	}
}

/*
 * bio_check_pages_dirty() will check that all the BIO's pages are still dirty.
 * If they are, then fine.  If, however, some pages are clean then they must
 * have been written out during the direct-IO read.  So we take another ref on
 * the BIO and the offending pages and re-dirty the pages in process context.
 *
 * It is expected that bio_check_pages_dirty() will wholly own the BIO from
 * here on.  It will run one page_cache_release() against each page and will
 * run one bio_put() against the BIO.
 */

1326
static void bio_dirty_fn(struct work_struct *work);
L
Linus Torvalds 已提交
1327

1328
static DECLARE_WORK(bio_dirty_work, bio_dirty_fn);
L
Linus Torvalds 已提交
1329 1330 1331 1332 1333 1334
static DEFINE_SPINLOCK(bio_dirty_lock);
static struct bio *bio_dirty_list;

/*
 * This runs in process context
 */
1335
static void bio_dirty_fn(struct work_struct *work)
L
Linus Torvalds 已提交
1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390
{
	unsigned long flags;
	struct bio *bio;

	spin_lock_irqsave(&bio_dirty_lock, flags);
	bio = bio_dirty_list;
	bio_dirty_list = NULL;
	spin_unlock_irqrestore(&bio_dirty_lock, flags);

	while (bio) {
		struct bio *next = bio->bi_private;

		bio_set_pages_dirty(bio);
		bio_release_pages(bio);
		bio_put(bio);
		bio = next;
	}
}

void bio_check_pages_dirty(struct bio *bio)
{
	struct bio_vec *bvec = bio->bi_io_vec;
	int nr_clean_pages = 0;
	int i;

	for (i = 0; i < bio->bi_vcnt; i++) {
		struct page *page = bvec[i].bv_page;

		if (PageDirty(page) || PageCompound(page)) {
			page_cache_release(page);
			bvec[i].bv_page = NULL;
		} else {
			nr_clean_pages++;
		}
	}

	if (nr_clean_pages) {
		unsigned long flags;

		spin_lock_irqsave(&bio_dirty_lock, flags);
		bio->bi_private = bio_dirty_list;
		bio_dirty_list = bio;
		spin_unlock_irqrestore(&bio_dirty_lock, flags);
		schedule_work(&bio_dirty_work);
	} else {
		bio_put(bio);
	}
}

/**
 * bio_endio - end I/O on a bio
 * @bio:	bio
 * @error:	error, if any
 *
 * Description:
1391
 *   bio_endio() will end I/O on the whole bio. bio_endio() is the
N
NeilBrown 已提交
1392 1393 1394 1395 1396 1397
 *   preferred way to end I/O on a bio, it takes care of clearing
 *   BIO_UPTODATE on error. @error is 0 on success, and and one of the
 *   established -Exxxx (-EIO, for instance) error values in case
 *   something went wrong. Noone should call bi_end_io() directly on a
 *   bio unless they own it and thus know that it has an end_io
 *   function.
L
Linus Torvalds 已提交
1398
 **/
1399
void bio_endio(struct bio *bio, int error)
L
Linus Torvalds 已提交
1400 1401 1402
{
	if (error)
		clear_bit(BIO_UPTODATE, &bio->bi_flags);
N
NeilBrown 已提交
1403 1404
	else if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
		error = -EIO;
L
Linus Torvalds 已提交
1405

N
NeilBrown 已提交
1406
	if (bio->bi_end_io)
1407
		bio->bi_end_io(bio, error);
L
Linus Torvalds 已提交
1408 1409 1410 1411 1412 1413 1414
}

void bio_pair_release(struct bio_pair *bp)
{
	if (atomic_dec_and_test(&bp->cnt)) {
		struct bio *master = bp->bio1.bi_private;

1415
		bio_endio(master, bp->error);
L
Linus Torvalds 已提交
1416 1417 1418 1419
		mempool_free(bp, bp->bio2.bi_private);
	}
}

1420
static void bio_pair_end_1(struct bio *bi, int err)
L
Linus Torvalds 已提交
1421 1422 1423 1424 1425 1426 1427 1428 1429
{
	struct bio_pair *bp = container_of(bi, struct bio_pair, bio1);

	if (err)
		bp->error = err;

	bio_pair_release(bp);
}

1430
static void bio_pair_end_2(struct bio *bi, int err)
L
Linus Torvalds 已提交
1431 1432 1433 1434 1435 1436 1437 1438 1439 1440
{
	struct bio_pair *bp = container_of(bi, struct bio_pair, bio2);

	if (err)
		bp->error = err;

	bio_pair_release(bp);
}

/*
1441
 * split a bio - only worry about a bio with a single page in its iovec
L
Linus Torvalds 已提交
1442
 */
D
Denis ChengRq 已提交
1443
struct bio_pair *bio_split(struct bio *bi, int first_sectors)
L
Linus Torvalds 已提交
1444
{
D
Denis ChengRq 已提交
1445
	struct bio_pair *bp = mempool_alloc(bio_split_pool, GFP_NOIO);
L
Linus Torvalds 已提交
1446 1447 1448 1449

	if (!bp)
		return bp;

1450
	trace_block_split(bdev_get_queue(bi->bi_bdev), bi,
1451 1452
				bi->bi_sector + first_sectors);

L
Linus Torvalds 已提交
1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471
	BUG_ON(bi->bi_vcnt != 1);
	BUG_ON(bi->bi_idx != 0);
	atomic_set(&bp->cnt, 3);
	bp->error = 0;
	bp->bio1 = *bi;
	bp->bio2 = *bi;
	bp->bio2.bi_sector += first_sectors;
	bp->bio2.bi_size -= first_sectors << 9;
	bp->bio1.bi_size = first_sectors << 9;

	bp->bv1 = bi->bi_io_vec[0];
	bp->bv2 = bi->bi_io_vec[0];
	bp->bv2.bv_offset += first_sectors << 9;
	bp->bv2.bv_len -= first_sectors << 9;
	bp->bv1.bv_len = first_sectors << 9;

	bp->bio1.bi_io_vec = &bp->bv1;
	bp->bio2.bi_io_vec = &bp->bv2;

1472 1473 1474
	bp->bio1.bi_max_vecs = 1;
	bp->bio2.bi_max_vecs = 1;

L
Linus Torvalds 已提交
1475 1476 1477 1478
	bp->bio1.bi_end_io = bio_pair_end_1;
	bp->bio2.bi_end_io = bio_pair_end_2;

	bp->bio1.bi_private = bi;
D
Denis ChengRq 已提交
1479
	bp->bio2.bi_private = bio_split_pool;
L
Linus Torvalds 已提交
1480

1481 1482 1483
	if (bio_integrity(bi))
		bio_integrity_split(bi, bp, first_sectors);

L
Linus Torvalds 已提交
1484 1485 1486
	return bp;
}

1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522
/**
 *      bio_sector_offset - Find hardware sector offset in bio
 *      @bio:           bio to inspect
 *      @index:         bio_vec index
 *      @offset:        offset in bv_page
 *
 *      Return the number of hardware sectors between beginning of bio
 *      and an end point indicated by a bio_vec index and an offset
 *      within that vector's page.
 */
sector_t bio_sector_offset(struct bio *bio, unsigned short index,
			   unsigned int offset)
{
	unsigned int sector_sz = queue_hardsect_size(bio->bi_bdev->bd_disk->queue);
	struct bio_vec *bv;
	sector_t sectors;
	int i;

	sectors = 0;

	if (index >= bio->bi_idx)
		index = bio->bi_vcnt - 1;

	__bio_for_each_segment(bv, bio, i, 0) {
		if (i == index) {
			if (offset > bv->bv_offset)
				sectors += (offset - bv->bv_offset) / sector_sz;
			break;
		}

		sectors += bv->bv_len / sector_sz;
	}

	return sectors;
}
EXPORT_SYMBOL(bio_sector_offset);
L
Linus Torvalds 已提交
1523 1524 1525 1526 1527

/*
 * create memory pools for biovec's in a bio_set.
 * use the global biovec slabs created for general use.
 */
1528
static int biovec_create_pools(struct bio_set *bs, int pool_entries)
L
Linus Torvalds 已提交
1529
{
1530
	struct biovec_slab *bp = bvec_slabs + BIOVEC_MAX_IDX;
L
Linus Torvalds 已提交
1531

1532 1533 1534
	bs->bvec_pool = mempool_create_slab_pool(pool_entries, bp->slab);
	if (!bs->bvec_pool)
		return -ENOMEM;
L
Linus Torvalds 已提交
1535 1536 1537 1538 1539 1540

	return 0;
}

static void biovec_free_pools(struct bio_set *bs)
{
1541
	mempool_destroy(bs->bvec_pool);
L
Linus Torvalds 已提交
1542 1543 1544 1545 1546 1547 1548 1549
}

void bioset_free(struct bio_set *bs)
{
	if (bs->bio_pool)
		mempool_destroy(bs->bio_pool);

	biovec_free_pools(bs);
1550
	bio_put_slab(bs);
L
Linus Torvalds 已提交
1551 1552 1553 1554

	kfree(bs);
}

1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568
/**
 * bioset_create  - Create a bio_set
 * @pool_size:	Number of bio and bio_vecs to cache in the mempool
 * @front_pad:	Number of bytes to allocate in front of the returned bio
 *
 * Description:
 *    Set up a bio_set to be used with @bio_alloc_bioset. Allows the caller
 *    to ask for a number of bytes to be allocated in front of the bio.
 *    Front pad allocation is useful for embedding the bio inside
 *    another structure, to avoid allocating extra data to go with the bio.
 *    Note that the bio must be embedded at the END of that structure always,
 *    or things will break badly.
 */
struct bio_set *bioset_create(unsigned int pool_size, unsigned int front_pad)
L
Linus Torvalds 已提交
1569
{
1570
	unsigned int back_pad = BIO_INLINE_VECS * sizeof(struct bio_vec);
1571
	struct bio_set *bs;
L
Linus Torvalds 已提交
1572

1573
	bs = kzalloc(sizeof(*bs), GFP_KERNEL);
L
Linus Torvalds 已提交
1574 1575 1576
	if (!bs)
		return NULL;

1577
	bs->front_pad = front_pad;
1578

1579
	bs->bio_slab = bio_find_or_create_slab(front_pad + back_pad);
1580 1581 1582 1583 1584 1585
	if (!bs->bio_slab) {
		kfree(bs);
		return NULL;
	}

	bs->bio_pool = mempool_create_slab_pool(pool_size, bs->bio_slab);
L
Linus Torvalds 已提交
1586 1587 1588
	if (!bs->bio_pool)
		goto bad;

1589
	if (!biovec_create_pools(bs, pool_size))
L
Linus Torvalds 已提交
1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604
		return bs;

bad:
	bioset_free(bs);
	return NULL;
}

static void __init biovec_init_slabs(void)
{
	int i;

	for (i = 0; i < BIOVEC_NR_POOLS; i++) {
		int size;
		struct biovec_slab *bvs = bvec_slabs + i;

1605 1606 1607 1608 1609 1610 1611
#ifndef CONFIG_BLK_DEV_INTEGRITY
		if (bvs->nr_vecs <= BIO_INLINE_VECS) {
			bvs->slab = NULL;
			continue;
		}
#endif

L
Linus Torvalds 已提交
1612 1613
		size = bvs->nr_vecs * sizeof(struct bio_vec);
		bvs->slab = kmem_cache_create(bvs->name, size, 0,
1614
                                SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
L
Linus Torvalds 已提交
1615 1616 1617 1618 1619
	}
}

static int __init init_bio(void)
{
1620 1621 1622 1623 1624
	bio_slab_max = 2;
	bio_slab_nr = 0;
	bio_slabs = kzalloc(bio_slab_max * sizeof(struct bio_slab), GFP_KERNEL);
	if (!bio_slabs)
		panic("bio: can't allocate bios\n");
L
Linus Torvalds 已提交
1625 1626 1627

	biovec_init_slabs();

1628
	fs_bio_set = bioset_create(BIO_POOL_SIZE, 0);
L
Linus Torvalds 已提交
1629 1630 1631
	if (!fs_bio_set)
		panic("bio: can't allocate bios\n");

1632 1633
	bio_split_pool = mempool_create_kmalloc_pool(BIO_SPLIT_ENTRIES,
						     sizeof(struct bio_pair));
L
Linus Torvalds 已提交
1634 1635 1636 1637 1638 1639 1640 1641 1642
	if (!bio_split_pool)
		panic("bio: can't create split pool\n");

	return 0;
}

subsys_initcall(init_bio);

EXPORT_SYMBOL(bio_alloc);
J
Jens Axboe 已提交
1643
EXPORT_SYMBOL(bio_kmalloc);
L
Linus Torvalds 已提交
1644
EXPORT_SYMBOL(bio_put);
P
Peter Osterlund 已提交
1645
EXPORT_SYMBOL(bio_free);
L
Linus Torvalds 已提交
1646 1647 1648 1649 1650 1651
EXPORT_SYMBOL(bio_endio);
EXPORT_SYMBOL(bio_init);
EXPORT_SYMBOL(__bio_clone);
EXPORT_SYMBOL(bio_clone);
EXPORT_SYMBOL(bio_phys_segments);
EXPORT_SYMBOL(bio_add_page);
1652
EXPORT_SYMBOL(bio_add_pc_page);
L
Linus Torvalds 已提交
1653
EXPORT_SYMBOL(bio_get_nr_vecs);
J
Jens Axboe 已提交
1654 1655
EXPORT_SYMBOL(bio_map_user);
EXPORT_SYMBOL(bio_unmap_user);
M
Mike Christie 已提交
1656
EXPORT_SYMBOL(bio_map_kern);
1657
EXPORT_SYMBOL(bio_copy_kern);
L
Linus Torvalds 已提交
1658 1659 1660 1661 1662 1663 1664
EXPORT_SYMBOL(bio_pair_release);
EXPORT_SYMBOL(bio_split);
EXPORT_SYMBOL(bio_copy_user);
EXPORT_SYMBOL(bio_uncopy_user);
EXPORT_SYMBOL(bioset_create);
EXPORT_SYMBOL(bioset_free);
EXPORT_SYMBOL(bio_alloc_bioset);