zram_drv.c 34.5 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 18 19
#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt

#include <linux/module.h>
#include <linux/kernel.h>
20
#include <linux/bio.h>
21 22 23 24 25 26
#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>
27
#include <linux/slab.h>
28 29
#include <linux/string.h>
#include <linux/vmalloc.h>
30
#include <linux/err.h>
31
#include <linux/idr.h>
32
#include <linux/sysfs.h>
33

34
#include "zram_drv.h"
35

36
static DEFINE_IDR(zram_index_idr);
37 38 39
/* idr index must be protected */
static DEFINE_MUTEX(zram_index_mutex);

40
static int zram_major;
41
static const char *default_compressor = "lzo";
42 43

/* Module params (documentation at end) */
44
static unsigned int num_devices = 1;
45

46 47 48 49 50 51 52 53 54
static inline void deprecated_attr_warn(const char *name)
{
	pr_warn_once("%d (%s) Attribute %s (and others) will be removed. %s\n",
			task_pid_nr(current),
			current->comm,
			name,
			"See zram documentation.");
}

55
#define ZRAM_ATTR_RO(name)						\
56
static ssize_t name##_show(struct device *d,				\
57 58 59
				struct device_attribute *attr, char *b)	\
{									\
	struct zram *zram = dev_to_zram(d);				\
60 61
									\
	deprecated_attr_warn(__stringify(name));			\
62
	return scnprintf(b, PAGE_SIZE, "%llu\n",			\
63 64
		(u64)atomic64_read(&zram->stats.name));			\
}									\
65
static DEVICE_ATTR_RO(name);
66

67
static inline bool init_done(struct zram *zram)
68
{
69
	return zram->disksize;
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;
}

77
/* flag operations require table entry bit_spin_lock() being held */
78 79
static int zram_test_flag(struct zram_meta *meta, u32 index,
			enum zram_pageflags flag)
80
{
81 82
	return meta->table[index].value & BIT(flag);
}
83

84 85 86 87 88
static void zram_set_flag(struct zram_meta *meta, u32 index,
			enum zram_pageflags flag)
{
	meta->table[index].value |= BIT(flag);
}
89

90 91 92 93 94
static void zram_clear_flag(struct zram_meta *meta, u32 index,
			enum zram_pageflags flag)
{
	meta->table[index].value &= ~BIT(flag);
}
95

96 97 98
static size_t zram_get_obj_size(struct zram_meta *meta, u32 index)
{
	return meta->table[index].value & (BIT(ZRAM_FLAG_SHIFT) - 1);
99 100
}

101 102
static void zram_set_obj_size(struct zram_meta *meta,
					u32 index, size_t size)
103
{
104
	unsigned long flags = meta->table[index].value >> ZRAM_FLAG_SHIFT;
105

106 107 108
	meta->table[index].value = (flags << ZRAM_FLAG_SHIFT) | size;
}

109
static inline bool is_partial_io(struct bio_vec *bvec)
110 111 112 113 114 115 116
{
	return bvec->bv_len != PAGE_SIZE;
}

/*
 * Check if request is within bounds and aligned on zram logical blocks.
 */
117
static inline bool valid_io_request(struct zram *zram,
118 119 120 121 122 123
		sector_t start, unsigned int size)
{
	u64 end, bound;

	/* unaligned request */
	if (unlikely(start & (ZRAM_SECTOR_PER_LOGICAL_BLOCK - 1)))
124
		return false;
125
	if (unlikely(size & (ZRAM_LOGICAL_BLOCK_SIZE - 1)))
126
		return false;
127 128 129 130 131

	end = start + (size >> SECTOR_SHIFT);
	bound = zram->disksize >> SECTOR_SHIFT;
	/* out of range range */
	if (unlikely(start >= bound || end > bound || start > end))
132
		return false;
133 134

	/* I/O request is valid */
135
	return true;
136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
}

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;
}

static inline void update_used_max(struct zram *zram,
					const unsigned long pages)
{
	unsigned long old_max, cur_max;

	old_max = atomic_long_read(&zram->stats.max_used_pages);

	do {
		cur_max = old_max;
		if (pages > cur_max)
			old_max = atomic_long_cmpxchg(
				&zram->stats.max_used_pages, cur_max, pages);
	} while (old_max != cur_max);
}

160
static bool page_zero_filled(void *ptr)
161 162 163 164 165 166 167 168
{
	unsigned int pos;
	unsigned long *page;

	page = (unsigned long *)ptr;

	for (pos = 0; pos != PAGE_SIZE / sizeof(*page); pos++) {
		if (page[pos])
169
			return false;
170 171
	}

172
	return true;
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187
}

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);
188 189 190 191 192
}

static ssize_t initstate_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
193
	u32 val;
194 195
	struct zram *zram = dev_to_zram(dev);

196 197 198
	down_read(&zram->init_lock);
	val = init_done(zram);
	up_read(&zram->init_lock);
199

200
	return scnprintf(buf, PAGE_SIZE, "%u\n", val);
201 202
}

203 204 205 206 207 208 209 210
static ssize_t disksize_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct zram *zram = dev_to_zram(dev);

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

211 212 213 214 215
static ssize_t orig_data_size_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct zram *zram = dev_to_zram(dev);

216
	deprecated_attr_warn("orig_data_size");
217
	return scnprintf(buf, PAGE_SIZE, "%llu\n",
218
		(u64)(atomic64_read(&zram->stats.pages_stored)) << PAGE_SHIFT);
219 220 221 222 223 224 225 226
}

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);

227
	deprecated_attr_warn("mem_used_total");
228
	down_read(&zram->init_lock);
229 230
	if (init_done(zram)) {
		struct zram_meta *meta = zram->meta;
231
		val = zs_get_total_pages(meta->mem_pool);
232
	}
233 234
	up_read(&zram->init_lock);

235
	return scnprintf(buf, PAGE_SIZE, "%llu\n", val << PAGE_SHIFT);
236 237
}

238 239 240 241 242 243
static ssize_t mem_limit_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	u64 val;
	struct zram *zram = dev_to_zram(dev);

244
	deprecated_attr_warn("mem_limit");
245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269
	down_read(&zram->init_lock);
	val = zram->limit_pages;
	up_read(&zram->init_lock);

	return scnprintf(buf, PAGE_SIZE, "%llu\n", val << PAGE_SHIFT);
}

static ssize_t mem_limit_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
{
	u64 limit;
	char *tmp;
	struct zram *zram = dev_to_zram(dev);

	limit = memparse(buf, &tmp);
	if (buf == tmp) /* no chars parsed, invalid input */
		return -EINVAL;

	down_write(&zram->init_lock);
	zram->limit_pages = PAGE_ALIGN(limit) >> PAGE_SHIFT;
	up_write(&zram->init_lock);

	return len;
}

Minchan Kim's avatar
Minchan Kim committed
270 271 272 273 274 275
static ssize_t mem_used_max_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	u64 val = 0;
	struct zram *zram = dev_to_zram(dev);

276
	deprecated_attr_warn("mem_used_max");
Minchan Kim's avatar
Minchan Kim committed
277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296
	down_read(&zram->init_lock);
	if (init_done(zram))
		val = atomic_long_read(&zram->stats.max_used_pages);
	up_read(&zram->init_lock);

	return scnprintf(buf, PAGE_SIZE, "%llu\n", val << PAGE_SHIFT);
}

static ssize_t mem_used_max_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
{
	int err;
	unsigned long val;
	struct zram *zram = dev_to_zram(dev);

	err = kstrtoul(buf, 10, &val);
	if (err || val != 0)
		return -EINVAL;

	down_read(&zram->init_lock);
297 298
	if (init_done(zram)) {
		struct zram_meta *meta = zram->meta;
Minchan Kim's avatar
Minchan Kim committed
299 300
		atomic_long_set(&zram->stats.max_used_pages,
				zs_get_total_pages(meta->mem_pool));
301
	}
Minchan Kim's avatar
Minchan Kim committed
302 303 304 305 306
	up_read(&zram->init_lock);

	return len;
}

307 308 309 310 311 312 313 314 315
/*
 * We switched to per-cpu streams and this attr is not needed anymore.
 * However, we will keep it around for some time, because:
 * a) we may revert per-cpu streams in the future
 * b) it's visible to user space and we need to follow our 2 years
 *    retirement rule; but we already have a number of 'soon to be
 *    altered' attrs, so max_comp_streams need to wait for the next
 *    layoff cycle.
 */
316 317 318
static ssize_t max_comp_streams_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
319
	return scnprintf(buf, PAGE_SIZE, "%d\n", num_online_cpus());
320 321
}

322 323 324
static ssize_t max_comp_streams_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
{
325
	return len;
326 327
}

328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
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);
345 346
	size_t sz;

347 348 349
	if (!zcomp_available_algorithm(buf))
		return -EINVAL;

350 351 352 353 354 355 356
	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));
357 358 359 360 361 362

	/* ignore trailing newline */
	sz = strlen(zram->compressor);
	if (sz > 0 && zram->compressor[sz - 1] == '\n')
		zram->compressor[sz - 1] = 0x00;

363 364 365 366
	up_write(&zram->init_lock);
	return len;
}

367 368
static ssize_t compact_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t len)
369
{
370 371
	struct zram *zram = dev_to_zram(dev);
	struct zram_meta *meta;
372

373 374 375 376 377
	down_read(&zram->init_lock);
	if (!init_done(zram)) {
		up_read(&zram->init_lock);
		return -EINVAL;
	}
378

379
	meta = zram->meta;
380
	zs_compact(meta->mem_pool);
381
	up_read(&zram->init_lock);
382

383
	return len;
384 385
}

386 387
static ssize_t io_stat_show(struct device *dev,
		struct device_attribute *attr, char *buf)
388
{
389 390
	struct zram *zram = dev_to_zram(dev);
	ssize_t ret;
391

392 393 394 395 396 397 398 399
	down_read(&zram->init_lock);
	ret = scnprintf(buf, PAGE_SIZE,
			"%8llu %8llu %8llu %8llu\n",
			(u64)atomic64_read(&zram->stats.failed_reads),
			(u64)atomic64_read(&zram->stats.failed_writes),
			(u64)atomic64_read(&zram->stats.invalid_io),
			(u64)atomic64_read(&zram->stats.notify_free));
	up_read(&zram->init_lock);
400

401
	return ret;
402 403
}

404 405
static ssize_t mm_stat_show(struct device *dev,
		struct device_attribute *attr, char *buf)
406
{
407
	struct zram *zram = dev_to_zram(dev);
408
	struct zs_pool_stats pool_stats;
409 410 411
	u64 orig_size, mem_used = 0;
	long max_used;
	ssize_t ret;
412

413 414
	memset(&pool_stats, 0x00, sizeof(struct zs_pool_stats));

415
	down_read(&zram->init_lock);
416
	if (init_done(zram)) {
417
		mem_used = zs_get_total_pages(zram->meta->mem_pool);
418 419
		zs_pool_stats(zram->meta->mem_pool, &pool_stats);
	}
420

421 422
	orig_size = atomic64_read(&zram->stats.pages_stored);
	max_used = atomic_long_read(&zram->stats.max_used_pages);
423

424
	ret = scnprintf(buf, PAGE_SIZE,
425
			"%8llu %8llu %8llu %8lu %8ld %8llu %8lu\n",
426 427 428 429 430 431
			orig_size << PAGE_SHIFT,
			(u64)atomic64_read(&zram->stats.compr_data_size),
			mem_used << PAGE_SHIFT,
			zram->limit_pages << PAGE_SHIFT,
			max_used << PAGE_SHIFT,
			(u64)atomic64_read(&zram->stats.zero_pages),
432
			pool_stats.pages_compacted);
433
	up_read(&zram->init_lock);
434

435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
	return ret;
}

static DEVICE_ATTR_RO(io_stat);
static DEVICE_ATTR_RO(mm_stat);
ZRAM_ATTR_RO(num_reads);
ZRAM_ATTR_RO(num_writes);
ZRAM_ATTR_RO(failed_reads);
ZRAM_ATTR_RO(failed_writes);
ZRAM_ATTR_RO(invalid_io);
ZRAM_ATTR_RO(notify_free);
ZRAM_ATTR_RO(zero_pages);
ZRAM_ATTR_RO(compr_data_size);

static inline bool zram_meta_get(struct zram *zram)
{
	if (atomic_inc_not_zero(&zram->refcount))
		return true;
	return false;
}

static inline void zram_meta_put(struct zram *zram)
{
	atomic_dec(&zram->refcount);
}

static void zram_meta_free(struct zram_meta *meta, u64 disksize)
{
	size_t num_pages = disksize >> PAGE_SHIFT;
	size_t index;
465 466 467 468 469 470 471 472 473 474 475

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

		if (!handle)
			continue;

		zs_free(meta->mem_pool, handle);
	}

476 477 478 479 480
	zs_destroy_pool(meta->mem_pool);
	vfree(meta->table);
	kfree(meta);
}

481
static struct zram_meta *zram_meta_alloc(char *pool_name, u64 disksize)
482 483 484
{
	size_t num_pages;
	struct zram_meta *meta = kmalloc(sizeof(*meta), GFP_KERNEL);
485

486
	if (!meta)
487
		return NULL;
488 489 490 491 492

	num_pages = disksize >> PAGE_SHIFT;
	meta->table = vzalloc(num_pages * sizeof(*meta->table));
	if (!meta->table) {
		pr_err("Error allocating zram address table\n");
493
		goto out_error;
494 495
	}

496
	meta->mem_pool = zs_create_pool(pool_name);
497 498
	if (!meta->mem_pool) {
		pr_err("Error creating memory pool\n");
499
		goto out_error;
500 501 502 503
	}

	return meta;

504
out_error:
505 506
	vfree(meta->table);
	kfree(meta);
507
	return NULL;
508 509
}

510 511 512 513 514
/*
 * To protect concurrent access to the same index entry,
 * caller should hold this table index entry's bit_spinlock to
 * indicate this index entry is accessing.
 */
515
static void zram_free_page(struct zram *zram, size_t index)
516
{
Minchan Kim's avatar
Minchan Kim committed
517 518
	struct zram_meta *meta = zram->meta;
	unsigned long handle = meta->table[index].handle;
519

520
	if (unlikely(!handle)) {
521 522 523 524
		/*
		 * No memory is allocated for zero filled pages.
		 * Simply clear zero page flag.
		 */
Minchan Kim's avatar
Minchan Kim committed
525 526
		if (zram_test_flag(meta, index, ZRAM_ZERO)) {
			zram_clear_flag(meta, index, ZRAM_ZERO);
527
			atomic64_dec(&zram->stats.zero_pages);
528 529 530 531
		}
		return;
	}

Minchan Kim's avatar
Minchan Kim committed
532
	zs_free(meta->mem_pool, handle);
533

534 535
	atomic64_sub(zram_get_obj_size(meta, index),
			&zram->stats.compr_data_size);
536
	atomic64_dec(&zram->stats.pages_stored);
537

Minchan Kim's avatar
Minchan Kim committed
538
	meta->table[index].handle = 0;
539
	zram_set_obj_size(meta, index, 0);
540 541
}

542
static int zram_decompress_page(struct zram *zram, char *mem, u32 index)
543
{
544
	int ret = 0;
545
	unsigned char *cmem;
Minchan Kim's avatar
Minchan Kim committed
546
	struct zram_meta *meta = zram->meta;
Minchan Kim's avatar
Minchan Kim committed
547
	unsigned long handle;
Minchan Kim's avatar
Minchan Kim committed
548
	size_t size;
Minchan Kim's avatar
Minchan Kim committed
549

550
	bit_spin_lock(ZRAM_ACCESS, &meta->table[index].value);
Minchan Kim's avatar
Minchan Kim committed
551
	handle = meta->table[index].handle;
552
	size = zram_get_obj_size(meta, index);
553

Minchan Kim's avatar
Minchan Kim committed
554
	if (!handle || zram_test_flag(meta, index, ZRAM_ZERO)) {
555
		bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
556
		clear_page(mem);
557 558
		return 0;
	}
559

Minchan Kim's avatar
Minchan Kim committed
560
	cmem = zs_map_object(meta->mem_pool, handle, ZS_MM_RO);
Minchan Kim's avatar
Minchan Kim committed
561
	if (size == PAGE_SIZE)
562
		copy_page(mem, cmem);
563
	else
564
		ret = zcomp_decompress(zram->comp, cmem, size, mem);
Minchan Kim's avatar
Minchan Kim committed
565
	zs_unmap_object(meta->mem_pool, handle);
566
	bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
567

568
	/* Should NEVER happen. Return bio error if it does. */
569
	if (unlikely(ret)) {
570 571
		pr_err("Decompression failed! err=%d, page=%u\n", ret, index);
		return ret;
572
	}
573

574
	return 0;
575 576
}

577
static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
578
			  u32 index, int offset)
579 580
{
	int ret;
581 582
	struct page *page;
	unsigned char *user_mem, *uncmem = NULL;
Minchan Kim's avatar
Minchan Kim committed
583
	struct zram_meta *meta = zram->meta;
584 585
	page = bvec->bv_page;

586
	bit_spin_lock(ZRAM_ACCESS, &meta->table[index].value);
Minchan Kim's avatar
Minchan Kim committed
587 588
	if (unlikely(!meta->table[index].handle) ||
			zram_test_flag(meta, index, ZRAM_ZERO)) {
589
		bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
590
		handle_zero_page(bvec);
591 592
		return 0;
	}
593
	bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
594

595 596
	if (is_partial_io(bvec))
		/* Use  a temporary buffer to decompress the page */
597 598 599 600
		uncmem = kmalloc(PAGE_SIZE, GFP_NOIO);

	user_mem = kmap_atomic(page);
	if (!is_partial_io(bvec))
601 602 603
		uncmem = user_mem;

	if (!uncmem) {
604
		pr_err("Unable to allocate temp memory\n");
605 606 607
		ret = -ENOMEM;
		goto out_cleanup;
	}
608

609
	ret = zram_decompress_page(zram, uncmem, index);
610
	/* Should NEVER happen. Return bio error if it does. */
611
	if (unlikely(ret))
612
		goto out_cleanup;
613

614 615 616 617 618 619 620 621 622 623 624
	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;
625 626 627 628
}

static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
			   int offset)
629
{
630
	int ret = 0;
631
	size_t clen;
632
	unsigned long handle = 0;
633
	struct page *page;
634
	unsigned char *user_mem, *cmem, *src, *uncmem = NULL;
Minchan Kim's avatar
Minchan Kim committed
635
	struct zram_meta *meta = zram->meta;
636
	struct zcomp_strm *zstrm = NULL;
Minchan Kim's avatar
Minchan Kim committed
637
	unsigned long alloced_pages;
638

639
	page = bvec->bv_page;
640 641 642 643 644
	if (is_partial_io(bvec)) {
		/*
		 * This is a partial IO. We need to read the full page
		 * before to write the changes.
		 */
645
		uncmem = kmalloc(PAGE_SIZE, GFP_NOIO);
646 647 648 649
		if (!uncmem) {
			ret = -ENOMEM;
			goto out;
		}
650
		ret = zram_decompress_page(zram, uncmem, index);
651
		if (ret)
652 653 654
			goto out;
	}

655
compress_again:
656
	user_mem = kmap_atomic(page);
657
	if (is_partial_io(bvec)) {
658 659
		memcpy(uncmem + offset, user_mem + bvec->bv_offset,
		       bvec->bv_len);
660 661 662
		kunmap_atomic(user_mem);
		user_mem = NULL;
	} else {
663
		uncmem = user_mem;
664
	}
665 666

	if (page_zero_filled(uncmem)) {
667 668
		if (user_mem)
			kunmap_atomic(user_mem);
669
		/* Free memory associated with this sector now. */
670
		bit_spin_lock(ZRAM_ACCESS, &meta->table[index].value);
671
		zram_free_page(zram, index);
Minchan Kim's avatar
Minchan Kim committed
672
		zram_set_flag(meta, index, ZRAM_ZERO);
673
		bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
674

675
		atomic64_inc(&zram->stats.zero_pages);
676 677
		ret = 0;
		goto out;
678
	}
679

680
	zstrm = zcomp_strm_find(zram->comp);
681
	ret = zcomp_compress(zram->comp, zstrm, uncmem, &clen);
682 683 684 685 686
	if (!is_partial_io(bvec)) {
		kunmap_atomic(user_mem);
		user_mem = NULL;
		uncmem = NULL;
	}
687

688
	if (unlikely(ret)) {
689
		pr_err("Compression failed! err=%d\n", ret);
690
		goto out;
691
	}
692

693
	src = zstrm->buffer;
694 695
	if (unlikely(clen > max_zpage_size)) {
		clen = PAGE_SIZE;
696 697
		if (is_partial_io(bvec))
			src = uncmem;
698
	}
699

700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
	/*
	 * handle allocation has 2 paths:
	 * a) fast path is executed with preemption disabled (for
	 *  per-cpu streams) and has __GFP_DIRECT_RECLAIM bit clear,
	 *  since we can't sleep;
	 * b) slow path enables preemption and attempts to allocate
	 *  the page with __GFP_DIRECT_RECLAIM bit set. we have to
	 *  put per-cpu compression stream and, thus, to re-do
	 *  the compression once handle is allocated.
	 *
	 * if we have a 'non-null' handle here then we are coming
	 * from the slow path and handle has already been allocated.
	 */
	if (!handle)
		handle = zs_malloc(meta->mem_pool, clen,
				__GFP_KSWAPD_RECLAIM |
				__GFP_NOWARN |
				__GFP_HIGHMEM);
718
	if (!handle) {
719 720 721 722 723 724 725 726
		zcomp_strm_release(zram->comp, zstrm);
		zstrm = NULL;

		handle = zs_malloc(meta->mem_pool, clen,
				GFP_NOIO | __GFP_HIGHMEM);
		if (handle)
			goto compress_again;

727
		pr_err("Error allocating memory for compressed page: %u, size=%zu\n",
728
			index, clen);
729 730
		ret = -ENOMEM;
		goto out;
731
	}
732

Minchan Kim's avatar
Minchan Kim committed
733
	alloced_pages = zs_get_total_pages(meta->mem_pool);
734 735
	update_used_max(zram, alloced_pages);

Minchan Kim's avatar
Minchan Kim committed
736
	if (zram->limit_pages && alloced_pages > zram->limit_pages) {
737 738 739 740 741
		zs_free(meta->mem_pool, handle);
		ret = -ENOMEM;
		goto out;
	}

Minchan Kim's avatar
Minchan Kim committed
742
	cmem = zs_map_object(meta->mem_pool, handle, ZS_MM_WO);
743

744
	if ((clen == PAGE_SIZE) && !is_partial_io(bvec)) {
745
		src = kmap_atomic(page);
746
		copy_page(cmem, src);
747
		kunmap_atomic(src);
748 749 750
	} else {
		memcpy(cmem, src, clen);
	}
751

752
	zcomp_strm_release(zram->comp, zstrm);
753
	zstrm = NULL;
Minchan Kim's avatar
Minchan Kim committed
754
	zs_unmap_object(meta->mem_pool, handle);
755

756 757 758 759
	/*
	 * Free memory associated with this sector
	 * before overwriting unused sectors.
	 */
760
	bit_spin_lock(ZRAM_ACCESS, &meta->table[index].value);
761 762
	zram_free_page(zram, index);

Minchan Kim's avatar
Minchan Kim committed
763
	meta->table[index].handle = handle;
764 765
	zram_set_obj_size(meta, index, clen);
	bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
766

767
	/* Update stats */
768 769
	atomic64_add(clen, &zram->stats.compr_data_size);
	atomic64_inc(&zram->stats.pages_stored);
770
out:
771
	if (zstrm)
772
		zcomp_strm_release(zram->comp, zstrm);
773 774
	if (is_partial_io(bvec))
		kfree(uncmem);
775
	return ret;
776 777
}

Joonsoo Kim's avatar
Joonsoo Kim committed
778 779 780 781 782 783 784 785 786
/*
 * zram_bio_discard - handler on discard request
 * @index: physical block index in PAGE_SIZE units
 * @offset: byte offset within physical block
 */
static void zram_bio_discard(struct zram *zram, u32 index,
			     int offset, struct bio *bio)
{
	size_t n = bio->bi_iter.bi_size;
787
	struct zram_meta *meta = zram->meta;
Joonsoo Kim's avatar
Joonsoo Kim committed
788 789 790 791 792 793 794 795 796 797 798 799

	/*
	 * zram manages data in physical block size units. Because logical block
	 * size isn't identical with physical block size on some arch, we
	 * could get a discard request pointing to a specific offset within a
	 * certain physical block.  Although we can handle this request by
	 * reading that physiclal block and decompressing and partially zeroing
	 * and re-compressing and then re-storing it, this isn't reasonable
	 * because our intent with a discard request is to save memory.  So
	 * skipping this logical block is appropriate here.
	 */
	if (offset) {
800
		if (n <= (PAGE_SIZE - offset))
Joonsoo Kim's avatar
Joonsoo Kim committed
801 802
			return;

803
		n -= (PAGE_SIZE - offset);
Joonsoo Kim's avatar
Joonsoo Kim committed
804 805 806 807
		index++;
	}

	while (n >= PAGE_SIZE) {
808
		bit_spin_lock(ZRAM_ACCESS, &meta->table[index].value);
Joonsoo Kim's avatar
Joonsoo Kim committed
809
		zram_free_page(zram, index);
810
		bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
811
		atomic64_inc(&zram->stats.notify_free);
Joonsoo Kim's avatar
Joonsoo Kim committed
812 813 814 815 816
		index++;
		n -= PAGE_SIZE;
	}
}

817 818
static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index,
			int offset, int rw)
819
{
820
	unsigned long start_time = jiffies;
821 822
	int ret;

823 824
	generic_start_io_acct(rw, bvec->bv_len >> SECTOR_SHIFT,
			&zram->disk->part0);
825

826 827 828 829 830 831
	if (rw == READ) {
		atomic64_inc(&zram->stats.num_reads);
		ret = zram_bvec_read(zram, bvec, index, offset);
	} else {
		atomic64_inc(&zram->stats.num_writes);
		ret = zram_bvec_write(zram, bvec, index, offset);
832
	}
833

834
	generic_end_io_acct(rw, &zram->disk->part0, start_time);
835

836 837 838 839 840
	if (unlikely(ret)) {
		if (rw == READ)
			atomic64_inc(&zram->stats.failed_reads);
		else
			atomic64_inc(&zram->stats.failed_writes);
841
	}
842

843
	return ret;
844 845
}

846
static void __zram_make_request(struct zram *zram, struct bio *bio)
847
{
848
	int offset, rw;
849
	u32 index;
850 851
	struct bio_vec bvec;
	struct bvec_iter iter;
852

853 854 855
	index = bio->bi_iter.bi_sector >> SECTORS_PER_PAGE_SHIFT;
	offset = (bio->bi_iter.bi_sector &
		  (SECTORS_PER_PAGE - 1)) << SECTOR_SHIFT;
856

Joonsoo Kim's avatar
Joonsoo Kim committed
857 858
	if (unlikely(bio->bi_rw & REQ_DISCARD)) {
		zram_bio_discard(zram, index, offset, bio);
859
		bio_endio(bio);
Joonsoo Kim's avatar
Joonsoo Kim committed
860 861 862
		return;
	}

863
	rw = bio_data_dir(bio);
864
	bio_for_each_segment(bvec, bio, iter) {
865 866
		int max_transfer_size = PAGE_SIZE - offset;

867
		if (bvec.bv_len > max_transfer_size) {
868 869 870 871 872 873
			/*
			 * zram_bvec_rw() can only make operation on a single
			 * zram page. Split the bio vector.
			 */
			struct bio_vec bv;

874
			bv.bv_page = bvec.bv_page;
875
			bv.bv_len = max_transfer_size;
876
			bv.bv_offset = bvec.bv_offset;
877

878
			if (zram_bvec_rw(zram, &bv, index, offset, rw) < 0)
879 880
				goto out;

881
			bv.bv_len = bvec.bv_len - max_transfer_size;
882
			bv.bv_offset += max_transfer_size;
883
			if (zram_bvec_rw(zram, &bv, index + 1, 0, rw) < 0)
884 885
				goto out;
		} else
886
			if (zram_bvec_rw(zram, &bvec, index, offset, rw) < 0)
887 888
				goto out;

889
		update_position(&index, &offset, &bvec);
890
	}
891

892
	bio_endio(bio);
893
	return;
894 895 896 897 898 899

out:
	bio_io_error(bio);
}

/*
900
 * Handler function for all zram I/O requests.
901
 */
902
static blk_qc_t zram_make_request(struct request_queue *queue, struct bio *bio)
903
{
904
	struct zram *zram = queue->queuedata;
905

906
	if (unlikely(!zram_meta_get(zram)))
907
		goto error;
908

909 910
	blk_queue_split(queue, &bio, queue->bio_split);

911 912
	if (!valid_io_request(zram, bio->bi_iter.bi_sector,
					bio->bi_iter.bi_size)) {
913
		atomic64_inc(&zram->stats.invalid_io);
914
		goto put_zram;
915 916
	}

917
	__zram_make_request(zram, bio);
918
	zram_meta_put(zram);
919
	return BLK_QC_T_NONE;
920