zram_drv.c 21.5 KB
Newer Older
1
/*
2
 * Compressed RAM block device
3
 *
4
 * Copyright (C) 2008, 2009, 2010  Nitin Gupta
Minchan Kim's avatar
Minchan Kim committed
5
 *               2012, 2013 Minchan Kim
6 7 8 9 10 11 12 13 14
 *
 * This code is released using a dual license strategy: BSD/GPL
 * You can choose the licence that better fits your requirements.
 *
 * Released under the terms of 3-clause BSD License
 * Released under the terms of GNU General Public License Version 2.0
 *
 */

15
#define KMSG_COMPONENT "zram"
16 17
#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt

18 19 20 21
#ifdef CONFIG_ZRAM_DEBUG
#define DEBUG
#endif

22 23
#include <linux/module.h>
#include <linux/kernel.h>
24
#include <linux/bio.h>
25 26 27 28 29 30
#include <linux/bitops.h>
#include <linux/blkdev.h>
#include <linux/buffer_head.h>
#include <linux/device.h>
#include <linux/genhd.h>
#include <linux/highmem.h>
31
#include <linux/slab.h>
32 33 34
#include <linux/string.h>
#include <linux/vmalloc.h>

35
#include "zram_drv.h"
36 37

/* Globals */
38
static int zram_major;
39
static struct zram *zram_devices;
40
static const char *default_compressor = "lzo";
41 42

/* Module params (documentation at end) */
43
static unsigned int num_devices = 1;
44

45 46 47 48 49 50 51 52 53 54 55
#define ZRAM_ATTR_RO(name)						\
static ssize_t zram_attr_##name##_show(struct device *d,		\
				struct device_attribute *attr, char *b)	\
{									\
	struct zram *zram = dev_to_zram(d);				\
	return sprintf(b, "%llu\n",					\
		(u64)atomic64_read(&zram->stats.name));			\
}									\
static struct device_attribute dev_attr_##name =			\
	__ATTR(name, S_IRUGO, zram_attr_##name##_show, NULL);

56 57 58 59 60
static inline int init_done(struct zram *zram)
{
	return zram->meta != NULL;
}

61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
static inline struct zram *dev_to_zram(struct device *dev)
{
	return (struct zram *)dev_to_disk(dev)->private_data;
}

static ssize_t disksize_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct zram *zram = dev_to_zram(dev);

	return sprintf(buf, "%llu\n", zram->disksize);
}

static ssize_t initstate_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
77
	u32 val;
78 79
	struct zram *zram = dev_to_zram(dev);

80 81 82
	down_read(&zram->init_lock);
	val = init_done(zram);
	up_read(&zram->init_lock);
83

84
	return sprintf(buf, "%u\n", val);
85 86 87 88 89 90 91 92
}

static ssize_t orig_data_size_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct zram *zram = dev_to_zram(dev);

	return sprintf(buf, "%llu\n",
93
		(u64)(atomic64_read(&zram->stats.pages_stored)) << PAGE_SHIFT);
94 95 96 97 98 99 100 101 102 103
}

static ssize_t mem_used_total_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	u64 val = 0;
	struct zram *zram = dev_to_zram(dev);
	struct zram_meta *meta = zram->meta;

	down_read(&zram->init_lock);
104
	if (init_done(zram))
105 106 107 108 109 110
		val = zs_get_total_size_bytes(meta->mem_pool);
	up_read(&zram->init_lock);

	return sprintf(buf, "%llu\n", val);
}

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
static ssize_t max_comp_streams_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	int val;
	struct zram *zram = dev_to_zram(dev);

	down_read(&zram->init_lock);
	val = zram->max_comp_streams;
	up_read(&zram->init_lock);

	return sprintf(buf, "%d\n", val);
}

static ssize_t max_comp_streams_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
{
	int num;
	struct zram *zram = dev_to_zram(dev);

	if (kstrtoint(buf, 0, &num))
		return -EINVAL;
	if (num < 1)
		return -EINVAL;
	down_write(&zram->init_lock);
	if (init_done(zram)) {
136 137
		if (zcomp_set_max_streams(zram->comp, num))
			pr_info("Cannot change max compression streams\n");
138 139 140 141 142 143
	}
	zram->max_comp_streams = num;
	up_write(&zram->init_lock);
	return len;
}

144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
static ssize_t comp_algorithm_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	size_t sz;
	struct zram *zram = dev_to_zram(dev);

	down_read(&zram->init_lock);
	sz = zcomp_available_show(zram->compressor, buf);
	up_read(&zram->init_lock);

	return sz;
}

static ssize_t comp_algorithm_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
{
	struct zram *zram = dev_to_zram(dev);
	down_write(&zram->init_lock);
	if (init_done(zram)) {
		up_write(&zram->init_lock);
		pr_info("Can't change algorithm for initialized device\n");
		return -EBUSY;
	}
	strlcpy(zram->compressor, buf, sizeof(zram->compressor));
	up_write(&zram->init_lock);
	return len;
}

Minchan Kim's avatar
Minchan Kim committed
172
/* flag operations needs meta->tb_lock */
Minchan Kim's avatar
Minchan Kim committed
173
static int zram_test_flag(struct zram_meta *meta, u32 index,
174
			enum zram_pageflags flag)
175
{
Minchan Kim's avatar
Minchan Kim committed
176
	return meta->table[index].flags & BIT(flag);
177 178
}

Minchan Kim's avatar
Minchan Kim committed
179
static void zram_set_flag(struct zram_meta *meta, u32 index,
180
			enum zram_pageflags flag)
181
{
Minchan Kim's avatar
Minchan Kim committed
182
	meta->table[index].flags |= BIT(flag);
183 184
}

Minchan Kim's avatar
Minchan Kim committed
185
static void zram_clear_flag(struct zram_meta *meta, u32 index,
186
			enum zram_pageflags flag)
187
{
Minchan Kim's avatar
Minchan Kim committed
188
	meta->table[index].flags &= ~BIT(flag);
189 190
}

191 192 193 194 195 196 197 198 199 200 201
static inline int is_partial_io(struct bio_vec *bvec)
{
	return bvec->bv_len != PAGE_SIZE;
}

/*
 * Check if request is within bounds and aligned on zram logical blocks.
 */
static inline int valid_io_request(struct zram *zram, struct bio *bio)
{
	u64 start, end, bound;
202

203
	/* unaligned request */
204 205
	if (unlikely(bio->bi_iter.bi_sector &
		     (ZRAM_SECTOR_PER_LOGICAL_BLOCK - 1)))
206
		return 0;
207
	if (unlikely(bio->bi_iter.bi_size & (ZRAM_LOGICAL_BLOCK_SIZE - 1)))
208 209
		return 0;

210 211
	start = bio->bi_iter.bi_sector;
	end = start + (bio->bi_iter.bi_size >> SECTOR_SHIFT);
212 213
	bound = zram->disksize >> SECTOR_SHIFT;
	/* out of range range */
214
	if (unlikely(start >= bound || end > bound || start > end))
215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238
		return 0;

	/* I/O request is valid */
	return 1;
}

static void zram_meta_free(struct zram_meta *meta)
{
	zs_destroy_pool(meta->mem_pool);
	vfree(meta->table);
	kfree(meta);
}

static struct zram_meta *zram_meta_alloc(u64 disksize)
{
	size_t num_pages;
	struct zram_meta *meta = kmalloc(sizeof(*meta), GFP_KERNEL);
	if (!meta)
		goto out;

	num_pages = disksize >> PAGE_SHIFT;
	meta->table = vzalloc(num_pages * sizeof(*meta->table));
	if (!meta->table) {
		pr_err("Error allocating zram address table\n");
239
		goto free_meta;
240 241 242 243 244 245 246 247
	}

	meta->mem_pool = zs_create_pool(GFP_NOIO | __GFP_HIGHMEM);
	if (!meta->mem_pool) {
		pr_err("Error creating memory pool\n");
		goto free_table;
	}

Minchan Kim's avatar
Minchan Kim committed
248
	rwlock_init(&meta->tb_lock);
249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266
	return meta;

free_table:
	vfree(meta->table);
free_meta:
	kfree(meta);
	meta = NULL;
out:
	return meta;
}

static void update_position(u32 *index, int *offset, struct bio_vec *bvec)
{
	if (*offset + bvec->bv_len >= PAGE_SIZE)
		(*index)++;
	*offset = (*offset + bvec->bv_len) % PAGE_SIZE;
}

267 268 269 270 271 272 273 274 275 276 277 278 279 280 281
static int page_zero_filled(void *ptr)
{
	unsigned int pos;
	unsigned long *page;

	page = (unsigned long *)ptr;

	for (pos = 0; pos != PAGE_SIZE / sizeof(*page); pos++) {
		if (page[pos])
			return 0;
	}

	return 1;
}

282 283 284 285 286 287 288 289 290 291 292 293 294 295 296
static void handle_zero_page(struct bio_vec *bvec)
{
	struct page *page = bvec->bv_page;
	void *user_mem;

	user_mem = kmap_atomic(page);
	if (is_partial_io(bvec))
		memset(user_mem + bvec->bv_offset, 0, bvec->bv_len);
	else
		clear_page(user_mem);
	kunmap_atomic(user_mem);

	flush_dcache_page(page);
}

Minchan Kim's avatar
Minchan Kim committed
297
/* NOTE: caller should hold meta->tb_lock with write-side */
298
static void zram_free_page(struct zram *zram, size_t index)
299
{
Minchan Kim's avatar
Minchan Kim committed
300 301
	struct zram_meta *meta = zram->meta;
	unsigned long handle = meta->table[index].handle;
302

303
	if (unlikely(!handle)) {
304 305 306 307
		/*
		 * No memory is allocated for zero filled pages.
		 * Simply clear zero page flag.
		 */
Minchan Kim's avatar
Minchan Kim committed
308 309
		if (zram_test_flag(meta, index, ZRAM_ZERO)) {
			zram_clear_flag(meta, index, ZRAM_ZERO);
310
			atomic64_dec(&zram->stats.zero_pages);
311 312 313 314
		}
		return;
	}

Minchan Kim's avatar
Minchan Kim committed
315
	zs_free(meta->mem_pool, handle);
316

317 318
	atomic64_sub(meta->table[index].size, &zram->stats.compr_data_size);
	atomic64_dec(&zram->stats.pages_stored);
319

Minchan Kim's avatar
Minchan Kim committed
320 321
	meta->table[index].handle = 0;
	meta->table[index].size = 0;
322 323
}

324
static int zram_decompress_page(struct zram *zram, char *mem, u32 index)
325
{
326
	int ret = 0;
327
	unsigned char *cmem;
Minchan Kim's avatar
Minchan Kim committed
328
	struct zram_meta *meta = zram->meta;
Minchan Kim's avatar
Minchan Kim committed
329 330 331 332 333 334
	unsigned long handle;
	u16 size;

	read_lock(&meta->tb_lock);
	handle = meta->table[index].handle;
	size = meta->table[index].size;
335

Minchan Kim's avatar
Minchan Kim committed
336
	if (!handle || zram_test_flag(meta, index, ZRAM_ZERO)) {
Minchan Kim's avatar
Minchan Kim committed
337
		read_unlock(&meta->tb_lock);
338
		clear_page(mem);
339 340
		return 0;
	}
341

Minchan Kim's avatar
Minchan Kim committed
342
	cmem = zs_map_object(meta->mem_pool, handle, ZS_MM_RO);
Minchan Kim's avatar
Minchan Kim committed
343
	if (size == PAGE_SIZE)
344
		copy_page(mem, cmem);
345
	else
346
		ret = zcomp_decompress(zram->comp, cmem, size, mem);
Minchan Kim's avatar
Minchan Kim committed
347
	zs_unmap_object(meta->mem_pool, handle);
Minchan Kim's avatar
Minchan Kim committed
348
	read_unlock(&meta->tb_lock);
349

350
	/* Should NEVER happen. Return bio error if it does. */
351
	if (unlikely(ret)) {
352
		pr_err("Decompression failed! err=%d, page=%u\n", ret, index);
353
		atomic64_inc(&zram->stats.failed_reads);
354
		return ret;
355
	}
356

357
	return 0;
358 359
}

360 361
static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
			  u32 index, int offset, struct bio *bio)
362 363
{
	int ret;
364 365
	struct page *page;
	unsigned char *user_mem, *uncmem = NULL;
Minchan Kim's avatar
Minchan Kim committed
366
	struct zram_meta *meta = zram->meta;
367 368
	page = bvec->bv_page;

Minchan Kim's avatar
Minchan Kim committed
369
	read_lock(&meta->tb_lock);
Minchan Kim's avatar
Minchan Kim committed
370 371
	if (unlikely(!meta->table[index].handle) ||
			zram_test_flag(meta, index, ZRAM_ZERO)) {
Minchan Kim's avatar
Minchan Kim committed
372
		read_unlock(&meta->tb_lock);
373
		handle_zero_page(bvec);
374 375
		return 0;
	}
Minchan Kim's avatar
Minchan Kim committed
376
	read_unlock(&meta->tb_lock);
377

378 379
	if (is_partial_io(bvec))
		/* Use  a temporary buffer to decompress the page */
380 381 382 383
		uncmem = kmalloc(PAGE_SIZE, GFP_NOIO);

	user_mem = kmap_atomic(page);
	if (!is_partial_io(bvec))
384 385 386 387 388 389 390
		uncmem = user_mem;

	if (!uncmem) {
		pr_info("Unable to allocate temp memory\n");
		ret = -ENOMEM;
		goto out_cleanup;
	}
391

392
	ret = zram_decompress_page(zram, uncmem, index);
393
	/* Should NEVER happen. Return bio error if it does. */
394
	if (unlikely(ret))
395
		goto out_cleanup;
396

397 398 399 400 401 402 403 404 405 406 407
	if (is_partial_io(bvec))
		memcpy(user_mem + bvec->bv_offset, uncmem + offset,
				bvec->bv_len);

	flush_dcache_page(page);
	ret = 0;
out_cleanup:
	kunmap_atomic(user_mem);
	if (is_partial_io(bvec))
		kfree(uncmem);
	return ret;
408 409 410 411
}

static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
			   int offset)
412
{
413
	int ret = 0;
414
	size_t clen;
415
	unsigned long handle;
416
	struct page *page;
417
	unsigned char *user_mem, *cmem, *src, *uncmem = NULL;
Minchan Kim's avatar
Minchan Kim committed
418
	struct zram_meta *meta = zram->meta;
419
	struct zcomp_strm *zstrm;
420
	bool locked = false;
421

422
	page = bvec->bv_page;
423 424 425 426 427
	if (is_partial_io(bvec)) {
		/*
		 * This is a partial IO. We need to read the full page
		 * before to write the changes.
		 */
428
		uncmem = kmalloc(PAGE_SIZE, GFP_NOIO);
429 430 431 432
		if (!uncmem) {
			ret = -ENOMEM;
			goto out;
		}
433
		ret = zram_decompress_page(zram, uncmem, index);
434
		if (ret)
435 436 437
			goto out;
	}

438
	zstrm = zcomp_strm_find(zram->comp);
439
	locked = true;
440
	user_mem = kmap_atomic(page);
441

442
	if (is_partial_io(bvec)) {
443 444
		memcpy(uncmem + offset, user_mem + bvec->bv_offset,
		       bvec->bv_len);
445 446 447
		kunmap_atomic(user_mem);
		user_mem = NULL;
	} else {
448
		uncmem = user_mem;
449
	}
450 451

	if (page_zero_filled(uncmem)) {
452
		kunmap_atomic(user_mem);
453
		/* Free memory associated with this sector now. */
Minchan Kim's avatar
Minchan Kim committed
454
		write_lock(&zram->meta->tb_lock);
455
		zram_free_page(zram, index);
Minchan Kim's avatar
Minchan Kim committed
456 457
		zram_set_flag(meta, index, ZRAM_ZERO);
		write_unlock(&zram->meta->tb_lock);
458

459
		atomic64_inc(&zram->stats.zero_pages);
460 461
		ret = 0;
		goto out;
462
	}
463

464
	ret = zcomp_compress(zram->comp, zstrm, uncmem, &clen);
465 466 467 468 469
	if (!is_partial_io(bvec)) {
		kunmap_atomic(user_mem);
		user_mem = NULL;
		uncmem = NULL;
	}
470

471
	if (unlikely(ret)) {
472
		pr_err("Compression failed! err=%d\n", ret);
473
		goto out;
474
	}
475
	src = zstrm->buffer;
476 477
	if (unlikely(clen > max_zpage_size)) {
		clen = PAGE_SIZE;
478 479
		if (is_partial_io(bvec))
			src = uncmem;
480
	}
481

Minchan Kim's avatar
Minchan Kim committed
482
	handle = zs_malloc(meta->mem_pool, clen);
483
	if (!handle) {
484 485
		pr_info("Error allocating memory for compressed page: %u, size=%zu\n",
			index, clen);
486 487
		ret = -ENOMEM;
		goto out;
488
	}
Minchan Kim's avatar
Minchan Kim committed
489
	cmem = zs_map_object(meta->mem_pool, handle, ZS_MM_WO);
490

491
	if ((clen == PAGE_SIZE) && !is_partial_io(bvec)) {
492
		src = kmap_atomic(page);
493
		copy_page(cmem, src);
494
		kunmap_atomic(src);
495 496 497
	} else {
		memcpy(cmem, src, clen);
	}
498

499 500
	zcomp_strm_release(zram->comp, zstrm);
	locked = false;
Minchan Kim's avatar
Minchan Kim committed
501
	zs_unmap_object(meta->mem_pool, handle);
502

503 504 505 506
	/*
	 * Free memory associated with this sector
	 * before overwriting unused sectors.
	 */
Minchan Kim's avatar
Minchan Kim committed
507
	write_lock(&zram->meta->tb_lock);
508 509
	zram_free_page(zram, index);

Minchan Kim's avatar
Minchan Kim committed
510 511
	meta->table[index].handle = handle;
	meta->table[index].size = clen;
Minchan Kim's avatar
Minchan Kim committed
512
	write_unlock(&zram->meta->tb_lock);
513

514
	/* Update stats */
515 516
	atomic64_add(clen, &zram->stats.compr_data_size);
	atomic64_inc(&zram->stats.pages_stored);
517
out:
518
	if (locked)
519
		zcomp_strm_release(zram->comp, zstrm);
520 521
	if (is_partial_io(bvec))
		kfree(uncmem);
522
	if (ret)
523
		atomic64_inc(&zram->stats.failed_writes);
524
	return ret;
525 526 527
}

static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index,
528
			int offset, struct bio *bio)
529
{
530
	int ret;
531
	int rw = bio_data_dir(bio);
532

533 534
	if (rw == READ) {
		atomic64_inc(&zram->stats.num_reads);
535
		ret = zram_bvec_read(zram, bvec, index, offset, bio);
536 537
	} else {
		atomic64_inc(&zram->stats.num_writes);
538
		ret = zram_bvec_write(zram, bvec, index, offset);
539
	}
540 541

	return ret;
542 543
}

Minchan Kim's avatar
Minchan Kim committed
544
static void zram_reset_device(struct zram *zram, bool reset_capacity)
545
{
546 547 548
	size_t index;
	struct zram_meta *meta;

549
	down_write(&zram->init_lock);
550
	if (!init_done(zram)) {
551
		up_write(&zram->init_lock);
552
		return;
553
	}
554 555 556 557 558 559 560 561 562 563 564

	meta = zram->meta;
	/* Free all pages that are still in this zram device */
	for (index = 0; index < zram->disksize >> PAGE_SHIFT; index++) {
		unsigned long handle = meta->table[index].handle;
		if (!handle)
			continue;

		zs_free(meta->mem_pool, handle);
	}

565
	zcomp_destroy(zram->comp);
566 567
	zram->max_comp_streams = 1;

568 569 570 571 572 573
	zram_meta_free(zram->meta);
	zram->meta = NULL;
	/* Reset stats */
	memset(&zram->stats, 0, sizeof(zram->stats));

	zram->disksize = 0;
Minchan Kim's avatar
Minchan Kim committed
574 575
	if (reset_capacity)
		set_capacity(zram->disk, 0);
576
	up_write(&zram->init_lock);
577 578 579 580 581 582
}

static ssize_t disksize_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
{
	u64 disksize;
583
	struct zcomp *comp;
584 585
	struct zram_meta *meta;
	struct zram *zram = dev_to_zram(dev);
586
	int err = -EINVAL;
587 588 589 590 591 592 593

	disksize = memparse(buf, NULL);
	if (!disksize)
		return -EINVAL;

	disksize = PAGE_ALIGN(disksize);
	meta = zram_meta_alloc(disksize);
594 595
	if (!meta)
		return -ENOMEM;
596

597 598 599 600 601 602 603
	comp = zcomp_create(zram->compressor, zram->max_comp_streams);
	if (!comp) {
		pr_info("Cannot initialise %s compressing backend\n",
				zram->compressor);
		goto out_cleanup;
	}

604
	down_write(&zram->init_lock);
605
	if (init_done(zram)) {
606
		up_write(&zram->init_lock);
607
		pr_info("Cannot change disksize for initialized device\n");
608
		err = -EBUSY;
609
		goto out_cleanup;
610 611
	}

612
	zram->meta = meta;
613
	zram->comp = comp;
614 615 616 617 618
	zram->disksize = disksize;
	set_capacity(zram->disk, zram->disksize >> SECTOR_SHIFT);
	up_write(&zram->init_lock);

	return len;
619

620 621 622
out_cleanup:
	if (comp)
		zcomp_destroy(comp);
623 624
	zram_meta_free(meta);
	return err;
625 626 627 628 629 630 631 632 633 634 635 636 637
}

static ssize_t reset_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
{
	int ret;
	unsigned short do_reset;
	struct zram *zram;
	struct block_device *bdev;

	zram = dev_to_zram(dev);
	bdev = bdget_disk(zram->disk, 0);

638 639 640
	if (!bdev)
		return -ENOMEM;

641
	/* Do not reset an active device! */
642 643 644 645
	if (bdev->bd_holders) {
		ret = -EBUSY;
		goto out;
	}
646 647 648

	ret = kstrtou16(buf, 10, &do_reset);
	if (ret)
649
		goto out;
650

651 652 653 654
	if (!do_reset) {
		ret = -EINVAL;
		goto out;
	}
655 656

	/* Make sure all pending I/O is finished */
657
	fsync_bdev(bdev);
658
	bdput(bdev);
659

Minchan Kim's avatar
Minchan Kim committed
660
	zram_reset_device(zram, true);
661
	return len;
662 663 664 665

out:
	bdput(bdev);
	return ret;
666 667
}

668
static void __zram_make_request(struct zram *zram, struct bio *bio)
669
{
670
	int offset;
671
	u32 index;
672 673
	struct bio_vec bvec;
	struct bvec_iter iter;
674

675 676 677
	index = bio->bi_iter.bi_sector >> SECTORS_PER_PAGE_SHIFT;
	offset = (bio->bi_iter.bi_sector &
		  (SECTORS_PER_PAGE - 1)) << SECTOR_SHIFT;
678

679
	bio_for_each_segment(bvec, bio, iter) {
680 681
		int max_transfer_size = PAGE_SIZE - offset;

682
		if (bvec.bv_len > max_transfer_size) {
683 684 685 686 687 688
			/*
			 * zram_bvec_rw() can only make operation on a single
			 * zram page. Split the bio vector.
			 */
			struct bio_vec bv;

689
			bv.bv_page = bvec.bv_page;
690
			bv.bv_len = max_transfer_size;
691
			bv.bv_offset = bvec.bv_offset;
692

693
			if (zram_bvec_rw(zram, &bv, index, offset, bio) < 0)
694 695
				goto out;

696
			bv.bv_len = bvec.bv_len - max_transfer_size;
697
			bv.bv_offset += max_transfer_size;
698
			if (zram_bvec_rw(zram, &bv, index + 1, 0, bio) < 0)
699 700
				goto out;
		} else
701
			if (zram_bvec_rw(zram, &bvec, index, offset, bio) < 0)
702 703
				goto out;

704
		update_position(&index, &offset, &bvec);
705
	}
706 707 708

	set_bit(BIO_UPTODATE, &bio->bi_flags);
	bio_endio(bio, 0);
709
	return;
710 711 712 713 714 715

out:
	bio_io_error(bio);
}

/*
716
 * Handler function for all zram I/O requests.
717
 */
718
static void zram_make_request(struct request_queue *queue, struct bio *bio)
719
{
720
	struct zram *zram = queue->queuedata;
721

722
	down_read(&zram->init_lock);
723
	if (unlikely(!init_done(zram)))
724
		goto error;
725

726
	if (!valid_io_request(zram, bio)) {
727
		atomic64_inc(&zram->stats.invalid_io);
728
		goto error;
729 730
	}

731
	__zram_make_request(zram, bio);
732
	up_read(&zram->init_lock);
733

734
	return;
735 736

error:
737
	up_read(&zram->init_lock);
738
	bio_io_error(bio);
739 740
}

Nitin Gupta's avatar
Nitin Gupta committed
741 742
static void zram_slot_free_notify(struct block_device *bdev,
				unsigned long index)
743
{
744
	struct zram *zram;
745
	struct zram_meta *meta;
746

747
	zram = bdev->bd_disk->private_data;
748
	meta = zram->meta;
749

750 751 752 753
	write_lock(&meta->tb_lock);
	zram_free_page(zram, index);
	write_unlock(&meta->tb_lock);
	atomic64_inc(&zram->stats.notify_free);
754 755
}

756 757
static const struct block_device_operations zram_devops = {
	.swap_slot_free_notify = zram_slot_free_notify,
758
	.owner = THIS_MODULE
759 760
};

761 762 763 764 765 766
static DEVICE_ATTR(disksize, S_IRUGO | S_IWUSR,
		disksize_show, disksize_store);
static DEVICE_ATTR(initstate, S_IRUGO, initstate_show, NULL);
static DEVICE_ATTR(reset, S_IWUSR, NULL, reset_store);
static DEVICE_ATTR(orig_data_size, S_IRUGO, orig_data_size_show, NULL);
static DEVICE_ATTR(mem_used_total, S_IRUGO, mem_used_total_show, NULL);
767 768
static DEVICE_ATTR(max_comp_streams, S_IRUGO | S_IWUSR,
		max_comp_streams_show, max_comp_streams_store);
769 770
static DEVICE_ATTR(comp_algorithm, S_IRUGO | S_IWUSR,
		comp_algorithm_show, comp_algorithm_store);
771

772 773
ZRAM_ATTR_RO(num_reads);
ZRAM_ATTR_RO(num_writes);
774 775
ZRAM_ATTR_RO(failed_reads);
ZRAM_ATTR_RO(failed_writes);
776 777 778 779 780
ZRAM_ATTR_RO(invalid_io);
ZRAM_ATTR_RO(notify_free);
ZRAM_ATTR_RO(zero_pages);
ZRAM_ATTR_RO(compr_data_size);

781 782 783 784 785 786
static struct attribute *zram_disk_attrs[] = {
	&dev_attr_disksize.attr,
	&dev_attr_initstate.attr,
	&dev_attr_reset.attr,
	&dev_attr_num_reads.attr,
	&dev_attr_num_writes.attr,
787 788
	&dev_attr_failed_reads.attr,
	&dev_attr_failed_writes.attr,
789 790 791 792 793 794
	&dev_attr_invalid_io.attr,
	&dev_attr_notify_free.attr,
	&dev_attr_zero_pages.attr,
	&dev_attr_orig_data_size.attr,
	&dev_attr_compr_data_size.attr,
	&dev_attr_mem_used_total.attr,
795
	&dev_attr_max_comp_streams.attr,
796
	&dev_attr_comp_algorithm.attr,
797 798 799 800 801 802 803
	NULL,
};

static struct attribute_group zram_disk_attr_group = {
	.attrs = zram_disk_attrs,
};

804
static int create_device(struct zram *zram, int device_id)
805
{
806
	int ret = -ENOMEM;
807

808
	init_rwsem(&zram->init_lock);
809

810 811
	zram->queue = blk_alloc_queue(GFP_KERNEL);
	if (!zram->queue) {
812 813
		pr_err("Error allocating disk queue for device %d\n",
			device_id);
814
		goto out;
815 816
	}

817 818
	blk_queue_make_request(zram->queue, zram_make_request);
	zram->queue->queuedata = zram;
819 820

	 /* gendisk structure */
821 822
	zram->disk = alloc_disk(1);
	if (!zram->disk) {
823
		pr_warn("Error allocating disk structure for device %d\n",
824
			device_id);
825
		goto out_free_queue;
826 827
	}

828 829 830 831 832 833
	zram->disk->major = zram_major;
	zram->disk->first_minor = device_id;
	zram->disk->fops = &zram_devops;
	zram->disk->queue = zram->queue;
	zram->disk->private_data = zram;
	snprintf(zram->disk->disk_name, 16, "zram%d", device_id);
834

835
	/* Actual capacity set using syfs (/sys/block/zram<id>/disksize */
836
	set_capacity(zram->disk, 0);
837 838
	/* zram devices sort of resembles non-rotational disks */
	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, zram->disk->queue);
839 840 841 842
	/*
	 * To ensure that we always get PAGE_SIZE aligned
	 * and n*PAGE_SIZED sized I/O requests.
	 */
843
	blk_queue_physical_block_size(zram->disk->queue, PAGE_SIZE);
844 845
	blk_queue_logical_block_size(zram->disk->queue,
					ZRAM_LOGICAL_BLOCK_SIZE);
846 847
	blk_queue_io_min(zram->disk->queue, PAGE_SIZE);
	blk_queue_io_opt(zram->disk->queue, PAGE_SIZE);
848

849
	add_disk(zram->disk);
850

851 852 853
	ret = sysfs_create_group(&disk_to_dev(zram->disk)->kobj,
				&zram_disk_attr_group);
	if (ret < 0) {
854
		pr_warn("Error creating sysfs group");
855
		goto out_free_disk;
856
	}
857
	strlcpy(zram->compressor, default_compressor, sizeof(zram->compressor));
858
	zram->meta = NULL;
859
	zram->max_comp_streams = 1;
860
	return 0;
861

862 863 864 865 866
out_free_disk:
	del_gendisk(zram->disk);
	put_disk(zram->disk);
out_free_queue:
	blk_cleanup_queue(zram->queue);
867 868
out:
	return ret;
869 870
}

871
static void destroy_device(struct zram *zram)
872
{
873 874 875
	sysfs_remove_group(&disk_to_dev(zram->disk)->kobj,
			&zram_disk_attr_group);

876 877
	del_gendisk(zram->disk);
	put_disk(zram->disk);
878

879
	blk_cleanup_queue(zram->queue);
880 881
}

882
static int __init zram_init(void)
883
{
884
	int ret, dev_id;
885

886
	if (num_devices > max_num_devices) {
887
		pr_warn("Invalid value for num_devices: %u\n",
888
				num_devices);
889 890
		ret = -EINVAL;
		goto out;
891 892
	}

893 894
	zram_major = register_blkdev(0, "zram");
	if (zram_major <= 0) {
895
		pr_warn("Unable to get major number\n");
896 897
		ret = -EBUSY;
		goto out;
898 899 900
	}

	/* Allocate the device array and initialize each one */
901
	zram_devices = kzalloc(num_devices * sizeof(struct zram), GFP_KERNEL);
902
	if (!zram_devices) {
903 904 905
		ret = -ENOMEM;
		goto unregister;
	}