zram_drv.c 19.6 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);
}

Minchan Kim's avatar
Minchan Kim committed
111
/* flag operations needs meta->tb_lock */
Minchan Kim's avatar
Minchan Kim committed
112
static int zram_test_flag(struct zram_meta *meta, u32 index,
113
			enum zram_pageflags flag)
114
{
Minchan Kim's avatar
Minchan Kim committed
115
	return meta->table[index].flags & BIT(flag);
116 117
}

Minchan Kim's avatar
Minchan Kim committed
118
static void zram_set_flag(struct zram_meta *meta, u32 index,
119
			enum zram_pageflags flag)
120
{
Minchan Kim's avatar
Minchan Kim committed
121
	meta->table[index].flags |= BIT(flag);
122 123
}

Minchan Kim's avatar
Minchan Kim committed
124
static void zram_clear_flag(struct zram_meta *meta, u32 index,
125
			enum zram_pageflags flag)
126
{
Minchan Kim's avatar
Minchan Kim committed
127
	meta->table[index].flags &= ~BIT(flag);
128 129
}

130 131 132 133 134 135 136 137 138 139 140
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;
141

142
	/* unaligned request */
143 144
	if (unlikely(bio->bi_iter.bi_sector &
		     (ZRAM_SECTOR_PER_LOGICAL_BLOCK - 1)))
145
		return 0;
146
	if (unlikely(bio->bi_iter.bi_size & (ZRAM_LOGICAL_BLOCK_SIZE - 1)))
147 148
		return 0;

149 150
	start = bio->bi_iter.bi_sector;
	end = start + (bio->bi_iter.bi_size >> SECTOR_SHIFT);
151 152
	bound = zram->disksize >> SECTOR_SHIFT;
	/* out of range range */
153
	if (unlikely(start >= bound || end > bound || start > end))
154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
		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");
178
		goto free_meta;
179 180 181 182 183 184 185 186
	}

	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
187
	rwlock_init(&meta->tb_lock);
188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205
	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;
}

206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
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;
}

221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
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
236
/* NOTE: caller should hold meta->tb_lock with write-side */
237
static void zram_free_page(struct zram *zram, size_t index)
238
{
Minchan Kim's avatar
Minchan Kim committed
239 240
	struct zram_meta *meta = zram->meta;
	unsigned long handle = meta->table[index].handle;
241

242
	if (unlikely(!handle)) {
243 244 245 246
		/*
		 * No memory is allocated for zero filled pages.
		 * Simply clear zero page flag.
		 */
Minchan Kim's avatar
Minchan Kim committed
247 248
		if (zram_test_flag(meta, index, ZRAM_ZERO)) {
			zram_clear_flag(meta, index, ZRAM_ZERO);
249
			atomic64_dec(&zram->stats.zero_pages);
250 251 252 253
		}
		return;
	}

Minchan Kim's avatar
Minchan Kim committed
254
	zs_free(meta->mem_pool, handle);
255

256 257
	atomic64_sub(meta->table[index].size, &zram->stats.compr_data_size);
	atomic64_dec(&zram->stats.pages_stored);
258

Minchan Kim's avatar
Minchan Kim committed
259 260
	meta->table[index].handle = 0;
	meta->table[index].size = 0;
261 262
}

263
static int zram_decompress_page(struct zram *zram, char *mem, u32 index)
264
{
265
	int ret = 0;
266
	unsigned char *cmem;
Minchan Kim's avatar
Minchan Kim committed
267
	struct zram_meta *meta = zram->meta;
Minchan Kim's avatar
Minchan Kim committed
268 269 270 271 272 273
	unsigned long handle;
	u16 size;

	read_lock(&meta->tb_lock);
	handle = meta->table[index].handle;
	size = meta->table[index].size;
274

Minchan Kim's avatar
Minchan Kim committed
275
	if (!handle || zram_test_flag(meta, index, ZRAM_ZERO)) {
Minchan Kim's avatar
Minchan Kim committed
276
		read_unlock(&meta->tb_lock);
277
		clear_page(mem);
278 279
		return 0;
	}
280

Minchan Kim's avatar
Minchan Kim committed
281
	cmem = zs_map_object(meta->mem_pool, handle, ZS_MM_RO);
Minchan Kim's avatar
Minchan Kim committed
282
	if (size == PAGE_SIZE)
283
		copy_page(mem, cmem);
284
	else
285
		ret = zcomp_decompress(zram->comp, cmem, size, mem);
Minchan Kim's avatar
Minchan Kim committed
286
	zs_unmap_object(meta->mem_pool, handle);
Minchan Kim's avatar
Minchan Kim committed
287
	read_unlock(&meta->tb_lock);
288

289
	/* Should NEVER happen. Return bio error if it does. */
290
	if (unlikely(ret)) {
291
		pr_err("Decompression failed! err=%d, page=%u\n", ret, index);
292
		atomic64_inc(&zram->stats.failed_reads);
293
		return ret;
294
	}
295

296
	return 0;
297 298
}

299 300
static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
			  u32 index, int offset, struct bio *bio)
301 302
{
	int ret;
303 304
	struct page *page;
	unsigned char *user_mem, *uncmem = NULL;
Minchan Kim's avatar
Minchan Kim committed
305
	struct zram_meta *meta = zram->meta;
306 307
	page = bvec->bv_page;

Minchan Kim's avatar
Minchan Kim committed
308
	read_lock(&meta->tb_lock);
Minchan Kim's avatar
Minchan Kim committed
309 310
	if (unlikely(!meta->table[index].handle) ||
			zram_test_flag(meta, index, ZRAM_ZERO)) {
Minchan Kim's avatar
Minchan Kim committed
311
		read_unlock(&meta->tb_lock);
312
		handle_zero_page(bvec);
313 314
		return 0;
	}
Minchan Kim's avatar
Minchan Kim committed
315
	read_unlock(&meta->tb_lock);
316

317 318
	if (is_partial_io(bvec))
		/* Use  a temporary buffer to decompress the page */
319 320 321 322
		uncmem = kmalloc(PAGE_SIZE, GFP_NOIO);

	user_mem = kmap_atomic(page);
	if (!is_partial_io(bvec))
323 324 325 326 327 328 329
		uncmem = user_mem;

	if (!uncmem) {
		pr_info("Unable to allocate temp memory\n");
		ret = -ENOMEM;
		goto out_cleanup;
	}
330

331
	ret = zram_decompress_page(zram, uncmem, index);
332
	/* Should NEVER happen. Return bio error if it does. */
333
	if (unlikely(ret))
334
		goto out_cleanup;
335

336 337 338 339 340 341 342 343 344 345 346
	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;
347 348 349 350
}

static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
			   int offset)
351
{
352
	int ret = 0;
353
	size_t clen;
354
	unsigned long handle;
355
	struct page *page;
356
	unsigned char *user_mem, *cmem, *src, *uncmem = NULL;
Minchan Kim's avatar
Minchan Kim committed
357
	struct zram_meta *meta = zram->meta;
358
	struct zcomp_strm *zstrm;
359
	bool locked = false;
360

361
	page = bvec->bv_page;
362 363 364 365 366
	if (is_partial_io(bvec)) {
		/*
		 * This is a partial IO. We need to read the full page
		 * before to write the changes.
		 */
367
		uncmem = kmalloc(PAGE_SIZE, GFP_NOIO);
368 369 370 371
		if (!uncmem) {
			ret = -ENOMEM;
			goto out;
		}
372
		ret = zram_decompress_page(zram, uncmem, index);
373
		if (ret)
374 375 376
			goto out;
	}

377
	zstrm = zcomp_strm_find(zram->comp);
378
	locked = true;
379
	user_mem = kmap_atomic(page);
380

381
	if (is_partial_io(bvec)) {
382 383
		memcpy(uncmem + offset, user_mem + bvec->bv_offset,
		       bvec->bv_len);
384 385 386
		kunmap_atomic(user_mem);
		user_mem = NULL;
	} else {
387
		uncmem = user_mem;
388
	}
389 390

	if (page_zero_filled(uncmem)) {
391
		kunmap_atomic(user_mem);
392
		/* Free memory associated with this sector now. */
Minchan Kim's avatar
Minchan Kim committed
393
		write_lock(&zram->meta->tb_lock);
394
		zram_free_page(zram, index);
Minchan Kim's avatar
Minchan Kim committed
395 396
		zram_set_flag(meta, index, ZRAM_ZERO);
		write_unlock(&zram->meta->tb_lock);
397

398
		atomic64_inc(&zram->stats.zero_pages);
399 400
		ret = 0;
		goto out;
401
	}
402

403
	ret = zcomp_compress(zram->comp, zstrm, uncmem, &clen);
404 405 406 407 408
	if (!is_partial_io(bvec)) {
		kunmap_atomic(user_mem);
		user_mem = NULL;
		uncmem = NULL;
	}
409

410
	if (unlikely(ret)) {
411
		pr_err("Compression failed! err=%d\n", ret);
412
		goto out;
413
	}
414
	src = zstrm->buffer;
415 416
	if (unlikely(clen > max_zpage_size)) {
		clen = PAGE_SIZE;
417 418
		if (is_partial_io(bvec))
			src = uncmem;
419
	}
420

Minchan Kim's avatar
Minchan Kim committed
421
	handle = zs_malloc(meta->mem_pool, clen);
422
	if (!handle) {
423 424
		pr_info("Error allocating memory for compressed page: %u, size=%zu\n",
			index, clen);
425 426
		ret = -ENOMEM;
		goto out;
427
	}
Minchan Kim's avatar
Minchan Kim committed
428
	cmem = zs_map_object(meta->mem_pool, handle, ZS_MM_WO);
429

430
	if ((clen == PAGE_SIZE) && !is_partial_io(bvec)) {
431
		src = kmap_atomic(page);
432
		copy_page(cmem, src);
433
		kunmap_atomic(src);
434 435 436
	} else {
		memcpy(cmem, src, clen);
	}
437

438 439
	zcomp_strm_release(zram->comp, zstrm);
	locked = false;
Minchan Kim's avatar
Minchan Kim committed
440
	zs_unmap_object(meta->mem_pool, handle);
441

442 443 444 445
	/*
	 * Free memory associated with this sector
	 * before overwriting unused sectors.
	 */
Minchan Kim's avatar
Minchan Kim committed
446
	write_lock(&zram->meta->tb_lock);
447 448
	zram_free_page(zram, index);

Minchan Kim's avatar
Minchan Kim committed
449 450
	meta->table[index].handle = handle;
	meta->table[index].size = clen;
Minchan Kim's avatar
Minchan Kim committed
451
	write_unlock(&zram->meta->tb_lock);
452

453
	/* Update stats */
454 455
	atomic64_add(clen, &zram->stats.compr_data_size);
	atomic64_inc(&zram->stats.pages_stored);
456
out:
457
	if (locked)
458
		zcomp_strm_release(zram->comp, zstrm);
459 460
	if (is_partial_io(bvec))
		kfree(uncmem);
461
	if (ret)
462
		atomic64_inc(&zram->stats.failed_writes);
463
	return ret;
464 465 466
}

static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index,
467
			int offset, struct bio *bio)
468
{
469
	int ret;
470
	int rw = bio_data_dir(bio);
471

472 473
	if (rw == READ) {
		atomic64_inc(&zram->stats.num_reads);
474
		ret = zram_bvec_read(zram, bvec, index, offset, bio);
475 476
	} else {
		atomic64_inc(&zram->stats.num_writes);
477
		ret = zram_bvec_write(zram, bvec, index, offset);
478
	}
479 480

	return ret;
481 482
}

Minchan Kim's avatar
Minchan Kim committed
483
static void zram_reset_device(struct zram *zram, bool reset_capacity)
484
{
485 486 487
	size_t index;
	struct zram_meta *meta;

488
	down_write(&zram->init_lock);
489
	if (!init_done(zram)) {
490
		up_write(&zram->init_lock);
491
		return;
492
	}
493 494 495 496 497 498 499 500 501 502 503

	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);
	}

504
	zcomp_destroy(zram->comp);
505 506 507 508 509 510
	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
511 512
	if (reset_capacity)
		set_capacity(zram->disk, 0);
513
	up_write(&zram->init_lock);
514 515 516 517 518 519 520 521
}

static ssize_t disksize_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
{
	u64 disksize;
	struct zram_meta *meta;
	struct zram *zram = dev_to_zram(dev);
522
	int err;
523 524 525 526 527 528 529

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

	disksize = PAGE_ALIGN(disksize);
	meta = zram_meta_alloc(disksize);
530 531
	if (!meta)
		return -ENOMEM;
532

533
	down_write(&zram->init_lock);
534
	if (init_done(zram)) {
535
		pr_info("Cannot change disksize for initialized device\n");
536 537 538 539 540 541 542 543 544 545
		err = -EBUSY;
		goto out_free_meta;
	}

	zram->comp = zcomp_create(default_compressor);
	if (!zram->comp) {
		pr_info("Cannot initialise %s compressing backend\n",
				default_compressor);
		err = -EINVAL;
		goto out_free_meta;
546 547
	}

548
	zram->meta = meta;
549 550 551 552 553
	zram->disksize = disksize;
	set_capacity(zram->disk, zram->disksize >> SECTOR_SHIFT);
	up_write(&zram->init_lock);

	return len;
554 555 556 557 558

out_free_meta:
	up_write(&zram->init_lock);
	zram_meta_free(meta);
	return err;
559 560 561 562 563 564 565 566 567 568 569 570 571
}

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);

572 573 574
	if (!bdev)
		return -ENOMEM;

575
	/* Do not reset an active device! */
576 577 578 579
	if (bdev->bd_holders) {
		ret = -EBUSY;
		goto out;
	}
580 581 582

	ret = kstrtou16(buf, 10, &do_reset);
	if (ret)
583
		goto out;
584

585 586 587 588
	if (!do_reset) {
		ret = -EINVAL;
		goto out;
	}
589 590

	/* Make sure all pending I/O is finished */
591
	fsync_bdev(bdev);
592
	bdput(bdev);
593

Minchan Kim's avatar
Minchan Kim committed
594
	zram_reset_device(zram, true);
595
	return len;
596 597 598 599

out:
	bdput(bdev);
	return ret;
600 601
}

602
static void __zram_make_request(struct zram *zram, struct bio *bio)
603
{
604
	int offset;
605
	u32 index;
606 607
	struct bio_vec bvec;
	struct bvec_iter iter;
608

609 610 611
	index = bio->bi_iter.bi_sector >> SECTORS_PER_PAGE_SHIFT;
	offset = (bio->bi_iter.bi_sector &
		  (SECTORS_PER_PAGE - 1)) << SECTOR_SHIFT;
612

613
	bio_for_each_segment(bvec, bio, iter) {
614 615
		int max_transfer_size = PAGE_SIZE - offset;

616
		if (bvec.bv_len > max_transfer_size) {
617 618 619 620 621 622
			/*
			 * zram_bvec_rw() can only make operation on a single
			 * zram page. Split the bio vector.
			 */
			struct bio_vec bv;

623
			bv.bv_page = bvec.bv_page;
624
			bv.bv_len = max_transfer_size;
625
			bv.bv_offset = bvec.bv_offset;
626

627
			if (zram_bvec_rw(zram, &bv, index, offset, bio) < 0)
628 629
				goto out;

630
			bv.bv_len = bvec.bv_len - max_transfer_size;
631
			bv.bv_offset += max_transfer_size;
632
			if (zram_bvec_rw(zram, &bv, index + 1, 0, bio) < 0)
633 634
				goto out;
		} else
635
			if (zram_bvec_rw(zram, &bvec, index, offset, bio) < 0)
636 637
				goto out;

638
		update_position(&index, &offset, &bvec);
639
	}
640 641 642

	set_bit(BIO_UPTODATE, &bio->bi_flags);
	bio_endio(bio, 0);
643
	return;
644 645 646 647 648 649

out:
	bio_io_error(bio);
}

/*
650
 * Handler function for all zram I/O requests.
651
 */
652
static void zram_make_request(struct request_queue *queue, struct bio *bio)
653
{
654
	struct zram *zram = queue->queuedata;
655

656
	down_read(&zram->init_lock);
657
	if (unlikely(!init_done(zram)))
658
		goto error;
659

660
	if (!valid_io_request(zram, bio)) {
661
		atomic64_inc(&zram->stats.invalid_io);
662
		goto error;
663 664
	}

665
	__zram_make_request(zram, bio);
666
	up_read(&zram->init_lock);
667

668
	return;
669 670

error:
671
	up_read(&zram->init_lock);
672
	bio_io_error(bio);
673 674
}

Nitin Gupta's avatar
Nitin Gupta committed
675 676
static void zram_slot_free_notify(struct block_device *bdev,
				unsigned long index)
677
{
678
	struct zram *zram;
679
	struct zram_meta *meta;
680

681
	zram = bdev->bd_disk->private_data;
682
	meta = zram->meta;
683

684 685 686 687
	write_lock(&meta->tb_lock);
	zram_free_page(zram, index);
	write_unlock(&meta->tb_lock);
	atomic64_inc(&zram->stats.notify_free);
688 689
}

690 691
static const struct block_device_operations zram_devops = {
	.swap_slot_free_notify = zram_slot_free_notify,
692
	.owner = THIS_MODULE
693 694
};

695 696 697 698 699 700 701
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);

702 703
ZRAM_ATTR_RO(num_reads);
ZRAM_ATTR_RO(num_writes);
704 705
ZRAM_ATTR_RO(failed_reads);
ZRAM_ATTR_RO(failed_writes);
706 707 708 709 710
ZRAM_ATTR_RO(invalid_io);
ZRAM_ATTR_RO(notify_free);
ZRAM_ATTR_RO(zero_pages);
ZRAM_ATTR_RO(compr_data_size);

711 712 713 714 715 716
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,
717 718
	&dev_attr_failed_reads.attr,
	&dev_attr_failed_writes.attr,
719 720 721 722 723 724 725 726 727 728 729 730 731
	&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,
	NULL,
};

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

732
static int create_device(struct zram *zram, int device_id)
733
{
734
	int ret = -ENOMEM;
735

736
	init_rwsem(&zram->init_lock);
737

738 739
	zram->queue = blk_alloc_queue(GFP_KERNEL);
	if (!zram->queue) {
740 741
		pr_err("Error allocating disk queue for device %d\n",
			device_id);
742
		goto out;
743 744
	}

745 746
	blk_queue_make_request(zram->queue, zram_make_request);
	zram->queue->queuedata = zram;
747 748

	 /* gendisk structure */
749 750
	zram->disk = alloc_disk(1);
	if (!zram->disk) {
751
		pr_warn("Error allocating disk structure for device %d\n",
752
			device_id);
753
		goto out_free_queue;
754 755
	}

756 757 758 759 760 761
	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);
762

763
	/* Actual capacity set using syfs (/sys/block/zram<id>/disksize */
764
	set_capacity(zram->disk, 0);
765 766
	/* zram devices sort of resembles non-rotational disks */
	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, zram->disk->queue);
767 768 769 770
	/*
	 * To ensure that we always get PAGE_SIZE aligned
	 * and n*PAGE_SIZED sized I/O requests.
	 */
771
	blk_queue_physical_block_size(zram->disk->queue, PAGE_SIZE);
772 773
	blk_queue_logical_block_size(zram->disk->queue,
					ZRAM_LOGICAL_BLOCK_SIZE);
774 775
	blk_queue_io_min(zram->disk->queue, PAGE_SIZE);
	blk_queue_io_opt(zram->disk->queue, PAGE_SIZE);
776

777
	add_disk(zram->disk);
778

779 780 781
	ret = sysfs_create_group(&disk_to_dev(zram->disk)->kobj,
				&zram_disk_attr_group);
	if (ret < 0) {
782
		pr_warn("Error creating sysfs group");
783
		goto out_free_disk;
784 785
	}

786
	zram->meta = NULL;
787
	return 0;
788

789 790 791 792 793
out_free_disk:
	del_gendisk(zram->disk);
	put_disk(zram->disk);
out_free_queue:
	blk_cleanup_queue(zram->queue);
794 795
out:
	return ret;
796 797
}

798
static void destroy_device(struct zram *zram)
799
{
800 801 802
	sysfs_remove_group(&disk_to_dev(zram->disk)->kobj,
			&zram_disk_attr_group);

803 804
	del_gendisk(zram->disk);
	put_disk(zram->disk);
805

806
	blk_cleanup_queue(zram->queue);
807 808
}

809
static int __init zram_init(void)
810
{
811
	int ret, dev_id;
812

813
	if (num_devices > max_num_devices) {
814
		pr_warn("Invalid value for num_devices: %u\n",
815
				num_devices);
816 817
		ret = -EINVAL;
		goto out;
818 819
	}

820 821
	zram_major = register_blkdev(0, "zram");
	if (zram_major <= 0) {
822
		pr_warn("Unable to get major number\n");
823 824
		ret = -EBUSY;
		goto out;
825 826 827
	}

	/* Allocate the device array and initialize each one */
828
	zram_devices = kzalloc(num_devices * sizeof(struct zram), GFP_KERNEL);
829
	if (!zram_devices) {
830 831 832
		ret = -ENOMEM;
		goto unregister;
	}
833

834
	for (dev_id = 0; dev_id < num_devices; dev_id++) {
835
		ret = create_device(&zram_devices[dev_id], dev_id);
836
		if (ret)
837
			goto free_devices;
838 839
	}

840 841
	pr_info("Created %u device(s) ...\n", num_devices);

842
	return 0;
843

844
free_devices:
845
	while (dev_id)
846 847
		destroy_device(&zram_devices[--dev_id]);
	kfree(zram_devices);
848
unregister:
849
	unregister_blkdev(zram_major, "zram");
850
out:
851 852 853
	return ret;
}

854
static void __exit zram_exit(void)
855 856
{
	int i;
857
	struct zram *zram;
858

859
	for (i = 0; i < num_devices; i++) {
860
		zram = &zram_devices[i];
861

862
		destroy_device(zram);
Minchan Kim's avatar
Minchan Kim committed
863 864 865 866 867
		/*
		 * Shouldn't access zram->disk after destroy_device
		 * because destroy_device already released zram->disk.
		 */
		zram_reset_device(zram, false);
868 869
	}

870
	unregister_blkdev(zram_major, "zram");
871

872
	kfree(zram_devices);
873 874 875
	pr_debug("Cleanup done!\n");
}

876 877
module_init(zram_init);
module_exit(zram_exit);