raid5-cache.c 55.4 KB
Newer Older
S
Shaohua Li 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2015 Shaohua Li <shli@fb.com>
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * This program is distributed in the hope 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.
 *
 */
#include <linux/kernel.h>
#include <linux/wait.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
#include <linux/raid/md_p.h>
S
Shaohua Li 已提交
19
#include <linux/crc32c.h>
S
Shaohua Li 已提交
20 21 22
#include <linux/random.h>
#include "md.h"
#include "raid5.h"
S
Song Liu 已提交
23
#include "bitmap.h"
S
Shaohua Li 已提交
24 25 26 27 28 29 30

/*
 * metadata/data stored in disk with 4k size unit (a block) regardless
 * underneath hardware sector size. only works with PAGE_SIZE == 4096
 */
#define BLOCK_SECTORS (8)

S
Shaohua Li 已提交
31
/*
32 33 34 35
 * log->max_free_space is min(1/4 disk size, 10G reclaimable space).
 *
 * In write through mode, the reclaim runs every log->max_free_space.
 * This can prevent the recovery scans for too long
S
Shaohua Li 已提交
36 37 38 39
 */
#define RECLAIM_MAX_FREE_SPACE (10 * 1024 * 1024 * 2) /* sector */
#define RECLAIM_MAX_FREE_SPACE_SHIFT (2)

40 41 42 43 44 45 46
/* wake up reclaim thread periodically */
#define R5C_RECLAIM_WAKEUP_INTERVAL (30 * HZ)
/* start flush with these full stripes */
#define R5C_FULL_STRIPE_FLUSH_BATCH 256
/* reclaim stripes in groups */
#define R5C_RECLAIM_STRIPE_GROUP (NR_STRIPE_HASH_LOCKS * 2)

C
Christoph Hellwig 已提交
47 48 49 50 51 52
/*
 * We only need 2 bios per I/O unit to make progress, but ensure we
 * have a few more available to not get too tight.
 */
#define R5L_POOL_SIZE	4

53 54 55 56 57 58 59 60 61 62
/*
 * r5c journal modes of the array: write-back or write-through.
 * write-through mode has identical behavior as existing log only
 * implementation.
 */
enum r5c_journal_mode {
	R5C_JOURNAL_MODE_WRITE_THROUGH = 0,
	R5C_JOURNAL_MODE_WRITE_BACK = 1,
};

S
Song Liu 已提交
63 64
static char *r5c_journal_mode_str[] = {"write-through",
				       "write-back"};
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
/*
 * raid5 cache state machine
 *
 * With rhe RAID cache, each stripe works in two phases:
 *	- caching phase
 *	- writing-out phase
 *
 * These two phases are controlled by bit STRIPE_R5C_CACHING:
 *   if STRIPE_R5C_CACHING == 0, the stripe is in writing-out phase
 *   if STRIPE_R5C_CACHING == 1, the stripe is in caching phase
 *
 * When there is no journal, or the journal is in write-through mode,
 * the stripe is always in writing-out phase.
 *
 * For write-back journal, the stripe is sent to caching phase on write
 * (r5c_try_caching_write). r5c_make_stripe_write_out() kicks off
 * the write-out phase by clearing STRIPE_R5C_CACHING.
 *
 * Stripes in caching phase do not write the raid disks. Instead, all
 * writes are committed from the log device. Therefore, a stripe in
 * caching phase handles writes as:
 *	- write to log device
 *	- return IO
 *
 * Stripes in writing-out phase handle writes as:
 *	- calculate parity
 *	- write pending data and parity to journal
 *	- write data and parity to raid disks
 *	- return IO for pending writes
 */

S
Shaohua Li 已提交
96 97 98 99 100 101 102
struct r5l_log {
	struct md_rdev *rdev;

	u32 uuid_checksum;

	sector_t device_size;		/* log device size, round to
					 * BLOCK_SECTORS */
S
Shaohua Li 已提交
103 104
	sector_t max_free_space;	/* reclaim run if free space is at
					 * this size */
S
Shaohua Li 已提交
105 106 107 108 109 110 111 112

	sector_t last_checkpoint;	/* log tail. where recovery scan
					 * starts from */
	u64 last_cp_seq;		/* log tail sequence */

	sector_t log_start;		/* log head. where new data appends */
	u64 seq;			/* log head sequence */

113 114 115
	sector_t next_checkpoint;
	u64 next_cp_seq;

S
Shaohua Li 已提交
116 117 118 119 120 121 122 123 124 125
	struct mutex io_mutex;
	struct r5l_io_unit *current_io;	/* current io_unit accepting new data */

	spinlock_t io_list_lock;
	struct list_head running_ios;	/* io_units which are still running,
					 * and have not yet been completely
					 * written to the log */
	struct list_head io_end_ios;	/* io_units which have been completely
					 * written to the log but not yet written
					 * to the RAID */
126 127
	struct list_head flushing_ios;	/* io_units which are waiting for log
					 * cache flush */
128
	struct list_head finished_ios;	/* io_units which settle down in log disk */
129
	struct bio flush_bio;
S
Shaohua Li 已提交
130

131 132
	struct list_head no_mem_stripes;   /* pending stripes, -ENOMEM */

S
Shaohua Li 已提交
133
	struct kmem_cache *io_kc;
134
	mempool_t *io_pool;
C
Christoph Hellwig 已提交
135
	struct bio_set *bs;
136
	mempool_t *meta_pool;
S
Shaohua Li 已提交
137

S
Shaohua Li 已提交
138 139 140 141 142 143 144 145
	struct md_thread *reclaim_thread;
	unsigned long reclaim_target;	/* number of space that need to be
					 * reclaimed.  if it's 0, reclaim spaces
					 * used by io_units which are in
					 * IO_UNIT_STRIPE_END state (eg, reclaim
					 * dones't wait for specific io_unit
					 * switching to IO_UNIT_STRIPE_END
					 * state) */
146
	wait_queue_head_t iounit_wait;
S
Shaohua Li 已提交
147

S
Shaohua Li 已提交
148 149
	struct list_head no_space_stripes; /* pending stripes, log has no space */
	spinlock_t no_space_stripes_lock;
150 151

	bool need_cache_flush;
152 153 154

	/* for r5c_cache */
	enum r5c_journal_mode r5c_journal_mode;
155 156 157 158 159 160

	/* all stripes in r5cache, in the order of seq at sh->log_start */
	struct list_head stripe_in_journal_list;

	spinlock_t stripe_in_journal_lock;
	atomic_t stripe_in_journal_count;
S
Shaohua Li 已提交
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185
};

/*
 * an IO range starts from a meta data block and end at the next meta data
 * block. The io unit's the meta data block tracks data/parity followed it. io
 * unit is written to log disk with normal write, as we always flush log disk
 * first and then start move data to raid disks, there is no requirement to
 * write io unit with FLUSH/FUA
 */
struct r5l_io_unit {
	struct r5l_log *log;

	struct page *meta_page;	/* store meta block */
	int meta_offset;	/* current offset in meta_page */

	struct bio *current_bio;/* current_bio accepting new data */

	atomic_t pending_stripe;/* how many stripes not flushed to raid */
	u64 seq;		/* seq number of the metablock */
	sector_t log_start;	/* where the io_unit starts */
	sector_t log_end;	/* where the io_unit ends */
	struct list_head log_sibling; /* log->running_ios */
	struct list_head stripe_list; /* stripes added to the io_unit */

	int state;
C
Christoph Hellwig 已提交
186
	bool need_split_bio;
S
Shaohua Li 已提交
187 188 189 190 191 192 193 194
};

/* r5l_io_unit state */
enum r5l_io_unit_state {
	IO_UNIT_RUNNING = 0,	/* accepting new IO */
	IO_UNIT_IO_START = 1,	/* io_unit bio start writing to log,
				 * don't accepting new bio */
	IO_UNIT_IO_END = 2,	/* io_unit bio finish writing to log */
195
	IO_UNIT_STRIPE_END = 3,	/* stripes data finished writing to raid */
S
Shaohua Li 已提交
196 197
};

198 199 200 201 202 203
bool r5c_is_writeback(struct r5l_log *log)
{
	return (log != NULL &&
		log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_BACK);
}

S
Shaohua Li 已提交
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238
static sector_t r5l_ring_add(struct r5l_log *log, sector_t start, sector_t inc)
{
	start += inc;
	if (start >= log->device_size)
		start = start - log->device_size;
	return start;
}

static sector_t r5l_ring_distance(struct r5l_log *log, sector_t start,
				  sector_t end)
{
	if (end >= start)
		return end - start;
	else
		return end + log->device_size - start;
}

static bool r5l_has_free_space(struct r5l_log *log, sector_t size)
{
	sector_t used_size;

	used_size = r5l_ring_distance(log, log->last_checkpoint,
					log->log_start);

	return log->device_size > used_size + size;
}

static void __r5l_set_io_unit_state(struct r5l_io_unit *io,
				    enum r5l_io_unit_state state)
{
	if (WARN_ON(io->state >= state))
		return;
	io->state = state;
}

S
Song Liu 已提交
239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275
static void
r5c_return_dev_pending_writes(struct r5conf *conf, struct r5dev *dev,
			      struct bio_list *return_bi)
{
	struct bio *wbi, *wbi2;

	wbi = dev->written;
	dev->written = NULL;
	while (wbi && wbi->bi_iter.bi_sector <
	       dev->sector + STRIPE_SECTORS) {
		wbi2 = r5_next_bio(wbi, dev->sector);
		if (!raid5_dec_bi_active_stripes(wbi)) {
			md_write_end(conf->mddev);
			bio_list_add(return_bi, wbi);
		}
		wbi = wbi2;
	}
}

void r5c_handle_cached_data_endio(struct r5conf *conf,
	  struct stripe_head *sh, int disks, struct bio_list *return_bi)
{
	int i;

	for (i = sh->disks; i--; ) {
		if (sh->dev[i].written) {
			set_bit(R5_UPTODATE, &sh->dev[i].flags);
			r5c_return_dev_pending_writes(conf, &sh->dev[i],
						      return_bi);
			bitmap_endwrite(conf->mddev->bitmap, sh->sector,
					STRIPE_SECTORS,
					!test_bit(STRIPE_DEGRADED, &sh->state),
					0);
		}
	}
}

276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373
/* Check whether we should flush some stripes to free up stripe cache */
void r5c_check_stripe_cache_usage(struct r5conf *conf)
{
	int total_cached;

	if (!r5c_is_writeback(conf->log))
		return;

	total_cached = atomic_read(&conf->r5c_cached_partial_stripes) +
		atomic_read(&conf->r5c_cached_full_stripes);

	/*
	 * The following condition is true for either of the following:
	 *   - stripe cache pressure high:
	 *          total_cached > 3/4 min_nr_stripes ||
	 *          empty_inactive_list_nr > 0
	 *   - stripe cache pressure moderate:
	 *          total_cached > 1/2 min_nr_stripes
	 */
	if (total_cached > conf->min_nr_stripes * 1 / 2 ||
	    atomic_read(&conf->empty_inactive_list_nr) > 0)
		r5l_wake_reclaim(conf->log, 0);
}

/*
 * flush cache when there are R5C_FULL_STRIPE_FLUSH_BATCH or more full
 * stripes in the cache
 */
void r5c_check_cached_full_stripe(struct r5conf *conf)
{
	if (!r5c_is_writeback(conf->log))
		return;

	/*
	 * wake up reclaim for R5C_FULL_STRIPE_FLUSH_BATCH cached stripes
	 * or a full stripe (chunk size / 4k stripes).
	 */
	if (atomic_read(&conf->r5c_cached_full_stripes) >=
	    min(R5C_FULL_STRIPE_FLUSH_BATCH,
		conf->chunk_sectors >> STRIPE_SHIFT))
		r5l_wake_reclaim(conf->log, 0);
}

/*
 * Total log space (in sectors) needed to flush all data in cache
 *
 * Currently, writing-out phase automatically includes all pending writes
 * to the same sector. So the reclaim of each stripe takes up to
 * (conf->raid_disks + 1) pages of log space.
 *
 * To totally avoid deadlock due to log space, the code reserves
 * (conf->raid_disks + 1) pages for each stripe in cache, which is not
 * necessary in most cases.
 *
 * To improve this, we will need writing-out phase to be able to NOT include
 * pending writes, which will reduce the requirement to
 * (conf->max_degraded + 1) pages per stripe in cache.
 */
static sector_t r5c_log_required_to_flush_cache(struct r5conf *conf)
{
	struct r5l_log *log = conf->log;

	if (!r5c_is_writeback(log))
		return 0;

	return BLOCK_SECTORS * (conf->raid_disks + 1) *
		atomic_read(&log->stripe_in_journal_count);
}

/*
 * evaluate log space usage and update R5C_LOG_TIGHT and R5C_LOG_CRITICAL
 *
 * R5C_LOG_TIGHT is set when free space on the log device is less than 3x of
 * reclaim_required_space. R5C_LOG_CRITICAL is set when free space on the log
 * device is less than 2x of reclaim_required_space.
 */
static inline void r5c_update_log_state(struct r5l_log *log)
{
	struct r5conf *conf = log->rdev->mddev->private;
	sector_t free_space;
	sector_t reclaim_space;

	if (!r5c_is_writeback(log))
		return;

	free_space = r5l_ring_distance(log, log->log_start,
				       log->last_checkpoint);
	reclaim_space = r5c_log_required_to_flush_cache(conf);
	if (free_space < 2 * reclaim_space)
		set_bit(R5C_LOG_CRITICAL, &conf->cache_state);
	else
		clear_bit(R5C_LOG_CRITICAL, &conf->cache_state);
	if (free_space < 3 * reclaim_space)
		set_bit(R5C_LOG_TIGHT, &conf->cache_state);
	else
		clear_bit(R5C_LOG_TIGHT, &conf->cache_state);
}

374 375 376 377
/*
 * Put the stripe into writing-out phase by clearing STRIPE_R5C_CACHING.
 * This function should only be called in write-back mode.
 */
378
void r5c_make_stripe_write_out(struct stripe_head *sh)
379 380 381 382 383 384 385 386
{
	struct r5conf *conf = sh->raid_conf;
	struct r5l_log *log = conf->log;

	BUG_ON(!r5c_is_writeback(log));

	WARN_ON(!test_bit(STRIPE_R5C_CACHING, &sh->state));
	clear_bit(STRIPE_R5C_CACHING, &sh->state);
S
Song Liu 已提交
387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424

	if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
		atomic_inc(&conf->preread_active_stripes);

	if (test_and_clear_bit(STRIPE_R5C_PARTIAL_STRIPE, &sh->state)) {
		BUG_ON(atomic_read(&conf->r5c_cached_partial_stripes) == 0);
		atomic_dec(&conf->r5c_cached_partial_stripes);
	}

	if (test_and_clear_bit(STRIPE_R5C_FULL_STRIPE, &sh->state)) {
		BUG_ON(atomic_read(&conf->r5c_cached_full_stripes) == 0);
		atomic_dec(&conf->r5c_cached_full_stripes);
	}
}

static void r5c_handle_data_cached(struct stripe_head *sh)
{
	int i;

	for (i = sh->disks; i--; )
		if (test_and_clear_bit(R5_Wantwrite, &sh->dev[i].flags)) {
			set_bit(R5_InJournal, &sh->dev[i].flags);
			clear_bit(R5_LOCKED, &sh->dev[i].flags);
		}
	clear_bit(STRIPE_LOG_TRAPPED, &sh->state);
}

/*
 * this journal write must contain full parity,
 * it may also contain some data pages
 */
static void r5c_handle_parity_cached(struct stripe_head *sh)
{
	int i;

	for (i = sh->disks; i--; )
		if (test_bit(R5_InJournal, &sh->dev[i].flags))
			set_bit(R5_Wantwrite, &sh->dev[i].flags);
425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443
}

/*
 * Setting proper flags after writing (or flushing) data and/or parity to the
 * log device. This is called from r5l_log_endio() or r5l_log_flush_endio().
 */
static void r5c_finish_cache_stripe(struct stripe_head *sh)
{
	struct r5l_log *log = sh->raid_conf->log;

	if (log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH) {
		BUG_ON(test_bit(STRIPE_R5C_CACHING, &sh->state));
		/*
		 * Set R5_InJournal for parity dev[pd_idx]. This means
		 * all data AND parity in the journal. For RAID 6, it is
		 * NOT necessary to set the flag for dev[qd_idx], as the
		 * two parities are written out together.
		 */
		set_bit(R5_InJournal, &sh->dev[sh->pd_idx].flags);
S
Song Liu 已提交
444 445 446 447 448 449
	} else if (test_bit(STRIPE_R5C_CACHING, &sh->state)) {
		r5c_handle_data_cached(sh);
	} else {
		r5c_handle_parity_cached(sh);
		set_bit(R5_InJournal, &sh->dev[sh->pd_idx].flags);
	}
450 451
}

452 453 454 455 456 457
static void r5l_io_run_stripes(struct r5l_io_unit *io)
{
	struct stripe_head *sh, *next;

	list_for_each_entry_safe(sh, next, &io->stripe_list, log_list) {
		list_del_init(&sh->log_list);
458 459 460

		r5c_finish_cache_stripe(sh);

461 462 463 464 465
		set_bit(STRIPE_HANDLE, &sh->state);
		raid5_release_stripe(sh);
	}
}

466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481
static void r5l_log_run_stripes(struct r5l_log *log)
{
	struct r5l_io_unit *io, *next;

	assert_spin_locked(&log->io_list_lock);

	list_for_each_entry_safe(io, next, &log->running_ios, log_sibling) {
		/* don't change list order */
		if (io->state < IO_UNIT_IO_END)
			break;

		list_move_tail(&io->log_sibling, &log->finished_ios);
		r5l_io_run_stripes(io);
	}
}

482 483 484 485 486 487 488 489 490 491 492 493 494 495
static void r5l_move_to_end_ios(struct r5l_log *log)
{
	struct r5l_io_unit *io, *next;

	assert_spin_locked(&log->io_list_lock);

	list_for_each_entry_safe(io, next, &log->running_ios, log_sibling) {
		/* don't change list order */
		if (io->state < IO_UNIT_IO_END)
			break;
		list_move_tail(&io->log_sibling, &log->io_end_ios);
	}
}

S
Shaohua Li 已提交
496 497 498 499
static void r5l_log_endio(struct bio *bio)
{
	struct r5l_io_unit *io = bio->bi_private;
	struct r5l_log *log = io->log;
500
	unsigned long flags;
S
Shaohua Li 已提交
501

S
Shaohua Li 已提交
502 503 504
	if (bio->bi_error)
		md_error(log->rdev->mddev, log->rdev);

S
Shaohua Li 已提交
505
	bio_put(bio);
506
	mempool_free(io->meta_page, log->meta_pool);
S
Shaohua Li 已提交
507

508 509
	spin_lock_irqsave(&log->io_list_lock, flags);
	__r5l_set_io_unit_state(io, IO_UNIT_IO_END);
510
	if (log->need_cache_flush)
511
		r5l_move_to_end_ios(log);
512 513
	else
		r5l_log_run_stripes(log);
514 515
	spin_unlock_irqrestore(&log->io_list_lock, flags);

516 517
	if (log->need_cache_flush)
		md_wakeup_thread(log->rdev->mddev->thread);
S
Shaohua Li 已提交
518 519 520 521 522 523
}

static void r5l_submit_current_io(struct r5l_log *log)
{
	struct r5l_io_unit *io = log->current_io;
	struct r5l_meta_block *block;
524
	unsigned long flags;
S
Shaohua Li 已提交
525 526 527 528 529 530 531
	u32 crc;

	if (!io)
		return;

	block = page_address(io->meta_page);
	block->meta_size = cpu_to_le32(io->meta_offset);
S
Shaohua Li 已提交
532
	crc = crc32c_le(log->uuid_checksum, block, PAGE_SIZE);
S
Shaohua Li 已提交
533 534 535
	block->checksum = cpu_to_le32(crc);

	log->current_io = NULL;
536 537 538
	spin_lock_irqsave(&log->io_list_lock, flags);
	__r5l_set_io_unit_state(io, IO_UNIT_IO_START);
	spin_unlock_irqrestore(&log->io_list_lock, flags);
S
Shaohua Li 已提交
539

540
	submit_bio(io->current_bio);
S
Shaohua Li 已提交
541 542
}

C
Christoph Hellwig 已提交
543
static struct bio *r5l_bio_alloc(struct r5l_log *log)
544
{
C
Christoph Hellwig 已提交
545
	struct bio *bio = bio_alloc_bioset(GFP_NOIO, BIO_MAX_PAGES, log->bs);
546

M
Mike Christie 已提交
547
	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
548
	bio->bi_bdev = log->rdev->bdev;
549
	bio->bi_iter.bi_sector = log->rdev->data_offset + log->log_start;
550 551 552 553

	return bio;
}

554 555 556 557
static void r5_reserve_log_entry(struct r5l_log *log, struct r5l_io_unit *io)
{
	log->log_start = r5l_ring_add(log, log->log_start, BLOCK_SECTORS);

558
	r5c_update_log_state(log);
559 560 561 562 563 564 565 566
	/*
	 * If we filled up the log device start from the beginning again,
	 * which will require a new bio.
	 *
	 * Note: for this to work properly the log size needs to me a multiple
	 * of BLOCK_SECTORS.
	 */
	if (log->log_start == 0)
C
Christoph Hellwig 已提交
567
		io->need_split_bio = true;
568 569 570 571

	io->log_end = log->log_start;
}

S
Shaohua Li 已提交
572 573 574 575 576
static struct r5l_io_unit *r5l_new_meta(struct r5l_log *log)
{
	struct r5l_io_unit *io;
	struct r5l_meta_block *block;

577 578 579 580 581
	io = mempool_alloc(log->io_pool, GFP_ATOMIC);
	if (!io)
		return NULL;
	memset(io, 0, sizeof(*io));

582 583 584 585
	io->log = log;
	INIT_LIST_HEAD(&io->log_sibling);
	INIT_LIST_HEAD(&io->stripe_list);
	io->state = IO_UNIT_RUNNING;
S
Shaohua Li 已提交
586

587
	io->meta_page = mempool_alloc(log->meta_pool, GFP_NOIO);
S
Shaohua Li 已提交
588
	block = page_address(io->meta_page);
589
	clear_page(block);
S
Shaohua Li 已提交
590 591 592 593 594 595 596
	block->magic = cpu_to_le32(R5LOG_MAGIC);
	block->version = R5LOG_VERSION;
	block->seq = cpu_to_le64(log->seq);
	block->position = cpu_to_le64(log->log_start);

	io->log_start = log->log_start;
	io->meta_offset = sizeof(struct r5l_meta_block);
597
	io->seq = log->seq++;
S
Shaohua Li 已提交
598

C
Christoph Hellwig 已提交
599 600 601
	io->current_bio = r5l_bio_alloc(log);
	io->current_bio->bi_end_io = r5l_log_endio;
	io->current_bio->bi_private = io;
602
	bio_add_page(io->current_bio, io->meta_page, PAGE_SIZE, 0);
S
Shaohua Li 已提交
603

604
	r5_reserve_log_entry(log, io);
S
Shaohua Li 已提交
605 606 607 608 609 610 611 612 613 614

	spin_lock_irq(&log->io_list_lock);
	list_add_tail(&io->log_sibling, &log->running_ios);
	spin_unlock_irq(&log->io_list_lock);

	return io;
}

static int r5l_get_meta(struct r5l_log *log, unsigned int payload_size)
{
615 616
	if (log->current_io &&
	    log->current_io->meta_offset + payload_size > PAGE_SIZE)
S
Shaohua Li 已提交
617 618
		r5l_submit_current_io(log);

619
	if (!log->current_io) {
620
		log->current_io = r5l_new_meta(log);
621 622 623 624
		if (!log->current_io)
			return -ENOMEM;
	}

S
Shaohua Li 已提交
625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653
	return 0;
}

static void r5l_append_payload_meta(struct r5l_log *log, u16 type,
				    sector_t location,
				    u32 checksum1, u32 checksum2,
				    bool checksum2_valid)
{
	struct r5l_io_unit *io = log->current_io;
	struct r5l_payload_data_parity *payload;

	payload = page_address(io->meta_page) + io->meta_offset;
	payload->header.type = cpu_to_le16(type);
	payload->header.flags = cpu_to_le16(0);
	payload->size = cpu_to_le32((1 + !!checksum2_valid) <<
				    (PAGE_SHIFT - 9));
	payload->location = cpu_to_le64(location);
	payload->checksum[0] = cpu_to_le32(checksum1);
	if (checksum2_valid)
		payload->checksum[1] = cpu_to_le32(checksum2);

	io->meta_offset += sizeof(struct r5l_payload_data_parity) +
		sizeof(__le32) * (1 + !!checksum2_valid);
}

static void r5l_append_payload_page(struct r5l_log *log, struct page *page)
{
	struct r5l_io_unit *io = log->current_io;

C
Christoph Hellwig 已提交
654 655
	if (io->need_split_bio) {
		struct bio *prev = io->current_bio;
656

C
Christoph Hellwig 已提交
657 658 659
		io->current_bio = r5l_bio_alloc(log);
		bio_chain(io->current_bio, prev);

660
		submit_bio(prev);
S
Shaohua Li 已提交
661 662
	}

C
Christoph Hellwig 已提交
663 664 665
	if (!bio_add_page(io->current_bio, page, PAGE_SIZE, 0))
		BUG();

666
	r5_reserve_log_entry(log, io);
S
Shaohua Li 已提交
667 668
}

669
static int r5l_log_stripe(struct r5l_log *log, struct stripe_head *sh,
S
Shaohua Li 已提交
670 671 672 673
			   int data_pages, int parity_pages)
{
	int i;
	int meta_size;
674
	int ret;
S
Shaohua Li 已提交
675 676 677 678 679 680 681 682
	struct r5l_io_unit *io;

	meta_size =
		((sizeof(struct r5l_payload_data_parity) + sizeof(__le32))
		 * data_pages) +
		sizeof(struct r5l_payload_data_parity) +
		sizeof(__le32) * parity_pages;

683 684 685 686
	ret = r5l_get_meta(log, meta_size);
	if (ret)
		return ret;

S
Shaohua Li 已提交
687 688 689
	io = log->current_io;

	for (i = 0; i < sh->disks; i++) {
S
Song Liu 已提交
690 691
		if (!test_bit(R5_Wantwrite, &sh->dev[i].flags) ||
		    test_bit(R5_InJournal, &sh->dev[i].flags))
S
Shaohua Li 已提交
692 693 694 695 696 697 698 699 700
			continue;
		if (i == sh->pd_idx || i == sh->qd_idx)
			continue;
		r5l_append_payload_meta(log, R5LOG_PAYLOAD_DATA,
					raid5_compute_blocknr(sh, i, 0),
					sh->dev[i].log_checksum, 0, false);
		r5l_append_payload_page(log, sh->dev[i].page);
	}

701
	if (parity_pages == 2) {
S
Shaohua Li 已提交
702 703 704 705 706
		r5l_append_payload_meta(log, R5LOG_PAYLOAD_PARITY,
					sh->sector, sh->dev[sh->pd_idx].log_checksum,
					sh->dev[sh->qd_idx].log_checksum, true);
		r5l_append_payload_page(log, sh->dev[sh->pd_idx].page);
		r5l_append_payload_page(log, sh->dev[sh->qd_idx].page);
707
	} else if (parity_pages == 1) {
S
Shaohua Li 已提交
708 709 710 711
		r5l_append_payload_meta(log, R5LOG_PAYLOAD_PARITY,
					sh->sector, sh->dev[sh->pd_idx].log_checksum,
					0, false);
		r5l_append_payload_page(log, sh->dev[sh->pd_idx].page);
712 713
	} else  /* Just writing data, not parity, in caching phase */
		BUG_ON(parity_pages != 0);
S
Shaohua Li 已提交
714 715 716 717

	list_add_tail(&sh->log_list, &io->stripe_list);
	atomic_inc(&io->pending_stripe);
	sh->log_io = io;
718

719 720 721 722 723 724 725 726 727 728 729 730
	if (log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH)
		return 0;

	if (sh->log_start == MaxSector) {
		BUG_ON(!list_empty(&sh->r5c));
		sh->log_start = io->log_start;
		spin_lock_irq(&log->stripe_in_journal_lock);
		list_add_tail(&sh->r5c,
			      &log->stripe_in_journal_list);
		spin_unlock_irq(&log->stripe_in_journal_lock);
		atomic_inc(&log->stripe_in_journal_count);
	}
731
	return 0;
S
Shaohua Li 已提交
732 733
}

734 735 736 737 738 739 740 741 742
/* add stripe to no_space_stripes, and then wake up reclaim */
static inline void r5l_add_no_space_stripe(struct r5l_log *log,
					   struct stripe_head *sh)
{
	spin_lock(&log->no_space_stripes_lock);
	list_add_tail(&sh->log_list, &log->no_space_stripes);
	spin_unlock(&log->no_space_stripes_lock);
}

S
Shaohua Li 已提交
743 744 745 746 747 748
/*
 * running in raid5d, where reclaim could wait for raid5d too (when it flushes
 * data from log to raid disks), so we shouldn't wait for reclaim here
 */
int r5l_write_stripe(struct r5l_log *log, struct stripe_head *sh)
{
749
	struct r5conf *conf = sh->raid_conf;
S
Shaohua Li 已提交
750 751 752 753
	int write_disks = 0;
	int data_pages, parity_pages;
	int reserve;
	int i;
754
	int ret = 0;
755
	bool wake_reclaim = false;
S
Shaohua Li 已提交
756 757 758 759 760 761 762 763 764 765 766

	if (!log)
		return -EAGAIN;
	/* Don't support stripe batch */
	if (sh->log_io || !test_bit(R5_Wantwrite, &sh->dev[sh->pd_idx].flags) ||
	    test_bit(STRIPE_SYNCING, &sh->state)) {
		/* the stripe is written to log, we start writing it to raid */
		clear_bit(STRIPE_LOG_TRAPPED, &sh->state);
		return -EAGAIN;
	}

767 768
	WARN_ON(test_bit(STRIPE_R5C_CACHING, &sh->state));

S
Shaohua Li 已提交
769 770 771
	for (i = 0; i < sh->disks; i++) {
		void *addr;

S
Song Liu 已提交
772 773
		if (!test_bit(R5_Wantwrite, &sh->dev[i].flags) ||
		    test_bit(R5_InJournal, &sh->dev[i].flags))
S
Shaohua Li 已提交
774
			continue;
S
Song Liu 已提交
775

S
Shaohua Li 已提交
776 777 778 779 780
		write_disks++;
		/* checksum is already calculated in last run */
		if (test_bit(STRIPE_LOG_TRAPPED, &sh->state))
			continue;
		addr = kmap_atomic(sh->dev[i].page);
S
Shaohua Li 已提交
781 782
		sh->dev[i].log_checksum = crc32c_le(log->uuid_checksum,
						    addr, PAGE_SIZE);
S
Shaohua Li 已提交
783 784 785 786 787 788
		kunmap_atomic(addr);
	}
	parity_pages = 1 + !!(sh->qd_idx >= 0);
	data_pages = write_disks - parity_pages;

	set_bit(STRIPE_LOG_TRAPPED, &sh->state);
789 790 791 792 793
	/*
	 * The stripe must enter state machine again to finish the write, so
	 * don't delay.
	 */
	clear_bit(STRIPE_DELAYED, &sh->state);
S
Shaohua Li 已提交
794 795 796 797 798 799
	atomic_inc(&sh->count);

	mutex_lock(&log->io_mutex);
	/* meta + data */
	reserve = (1 + write_disks) << (PAGE_SHIFT - 9);

800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835
	if (log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH) {
		if (!r5l_has_free_space(log, reserve)) {
			r5l_add_no_space_stripe(log, sh);
			wake_reclaim = true;
		} else {
			ret = r5l_log_stripe(log, sh, data_pages, parity_pages);
			if (ret) {
				spin_lock_irq(&log->io_list_lock);
				list_add_tail(&sh->log_list,
					      &log->no_mem_stripes);
				spin_unlock_irq(&log->io_list_lock);
			}
		}
	} else {  /* R5C_JOURNAL_MODE_WRITE_BACK */
		/*
		 * log space critical, do not process stripes that are
		 * not in cache yet (sh->log_start == MaxSector).
		 */
		if (test_bit(R5C_LOG_CRITICAL, &conf->cache_state) &&
		    sh->log_start == MaxSector) {
			r5l_add_no_space_stripe(log, sh);
			wake_reclaim = true;
			reserve = 0;
		} else if (!r5l_has_free_space(log, reserve)) {
			if (sh->log_start == log->last_checkpoint)
				BUG();
			else
				r5l_add_no_space_stripe(log, sh);
		} else {
			ret = r5l_log_stripe(log, sh, data_pages, parity_pages);
			if (ret) {
				spin_lock_irq(&log->io_list_lock);
				list_add_tail(&sh->log_list,
					      &log->no_mem_stripes);
				spin_unlock_irq(&log->io_list_lock);
			}
836
		}
S
Shaohua Li 已提交
837 838
	}

839
	mutex_unlock(&log->io_mutex);
840 841
	if (wake_reclaim)
		r5l_wake_reclaim(log, reserve);
S
Shaohua Li 已提交
842 843 844 845 846 847 848 849 850 851 852 853
	return 0;
}

void r5l_write_stripe_run(struct r5l_log *log)
{
	if (!log)
		return;
	mutex_lock(&log->io_mutex);
	r5l_submit_current_io(log);
	mutex_unlock(&log->io_mutex);
}

854 855 856 857 858 859 860 861 862 863 864 865 866 867
int r5l_handle_flush_request(struct r5l_log *log, struct bio *bio)
{
	if (!log)
		return -ENODEV;
	/*
	 * we flush log disk cache first, then write stripe data to raid disks.
	 * So if bio is finished, the log disk cache is flushed already. The
	 * recovery guarantees we can recovery the bio from log disk, so we
	 * don't need to flush again
	 */
	if (bio->bi_iter.bi_size == 0) {
		bio_endio(bio);
		return 0;
	}
J
Jens Axboe 已提交
868
	bio->bi_opf &= ~REQ_PREFLUSH;
869 870 871
	return -EAGAIN;
}

S
Shaohua Li 已提交
872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887
/* This will run after log space is reclaimed */
static void r5l_run_no_space_stripes(struct r5l_log *log)
{
	struct stripe_head *sh;

	spin_lock(&log->no_space_stripes_lock);
	while (!list_empty(&log->no_space_stripes)) {
		sh = list_first_entry(&log->no_space_stripes,
				      struct stripe_head, log_list);
		list_del_init(&sh->log_list);
		set_bit(STRIPE_HANDLE, &sh->state);
		raid5_release_stripe(sh);
	}
	spin_unlock(&log->no_space_stripes_lock);
}

888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915
/*
 * calculate new last_checkpoint
 * for write through mode, returns log->next_checkpoint
 * for write back, returns log_start of first sh in stripe_in_journal_list
 */
static sector_t r5c_calculate_new_cp(struct r5conf *conf)
{
	struct stripe_head *sh;
	struct r5l_log *log = conf->log;
	sector_t new_cp;
	unsigned long flags;

	if (log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH)
		return log->next_checkpoint;

	spin_lock_irqsave(&log->stripe_in_journal_lock, flags);
	if (list_empty(&conf->log->stripe_in_journal_list)) {
		/* all stripes flushed */
		spin_unlock(&log->stripe_in_journal_lock);
		return log->next_checkpoint;
	}
	sh = list_first_entry(&conf->log->stripe_in_journal_list,
			      struct stripe_head, r5c);
	new_cp = sh->log_start;
	spin_unlock_irqrestore(&log->stripe_in_journal_lock, flags);
	return new_cp;
}

916 917
static sector_t r5l_reclaimable_space(struct r5l_log *log)
{
918 919
	struct r5conf *conf = log->rdev->mddev->private;

920
	return r5l_ring_distance(log, log->last_checkpoint,
921
				 r5c_calculate_new_cp(conf));
922 923
}

924 925 926 927 928 929 930 931 932 933 934 935 936 937 938
static void r5l_run_no_mem_stripe(struct r5l_log *log)
{
	struct stripe_head *sh;

	assert_spin_locked(&log->io_list_lock);

	if (!list_empty(&log->no_mem_stripes)) {
		sh = list_first_entry(&log->no_mem_stripes,
				      struct stripe_head, log_list);
		list_del_init(&sh->log_list);
		set_bit(STRIPE_HANDLE, &sh->state);
		raid5_release_stripe(sh);
	}
}

939
static bool r5l_complete_finished_ios(struct r5l_log *log)
940 941 942 943 944 945
{
	struct r5l_io_unit *io, *next;
	bool found = false;

	assert_spin_locked(&log->io_list_lock);

946
	list_for_each_entry_safe(io, next, &log->finished_ios, log_sibling) {
947 948 949 950 951 952 953 954
		/* don't change list order */
		if (io->state < IO_UNIT_STRIPE_END)
			break;

		log->next_checkpoint = io->log_start;
		log->next_cp_seq = io->seq;

		list_del(&io->log_sibling);
955 956
		mempool_free(io, log->io_pool);
		r5l_run_no_mem_stripe(log);
957 958 959 960 961 962 963

		found = true;
	}

	return found;
}

964 965 966
static void __r5l_stripe_write_finished(struct r5l_io_unit *io)
{
	struct r5l_log *log = io->log;
967
	struct r5conf *conf = log->rdev->mddev->private;
968 969 970 971
	unsigned long flags;

	spin_lock_irqsave(&log->io_list_lock, flags);
	__r5l_set_io_unit_state(io, IO_UNIT_STRIPE_END);
972

973
	if (!r5l_complete_finished_ios(log)) {
974 975 976
		spin_unlock_irqrestore(&log->io_list_lock, flags);
		return;
	}
977

978 979
	if (r5l_reclaimable_space(log) > log->max_free_space ||
	    test_bit(R5C_LOG_TIGHT, &conf->cache_state))
980 981 982 983 984 985
		r5l_wake_reclaim(log, 0);

	spin_unlock_irqrestore(&log->io_list_lock, flags);
	wake_up(&log->iounit_wait);
}

S
Shaohua Li 已提交
986 987 988 989 990 991 992
void r5l_stripe_write_finished(struct stripe_head *sh)
{
	struct r5l_io_unit *io;

	io = sh->log_io;
	sh->log_io = NULL;

993 994
	if (io && atomic_dec_and_test(&io->pending_stripe))
		__r5l_stripe_write_finished(io);
S
Shaohua Li 已提交
995 996
}

997 998 999 1000 1001 1002 1003
static void r5l_log_flush_endio(struct bio *bio)
{
	struct r5l_log *log = container_of(bio, struct r5l_log,
		flush_bio);
	unsigned long flags;
	struct r5l_io_unit *io;

S
Shaohua Li 已提交
1004 1005 1006
	if (bio->bi_error)
		md_error(log->rdev->mddev, log->rdev);

1007
	spin_lock_irqsave(&log->io_list_lock, flags);
1008 1009
	list_for_each_entry(io, &log->flushing_ios, log_sibling)
		r5l_io_run_stripes(io);
1010
	list_splice_tail_init(&log->flushing_ios, &log->finished_ios);
1011 1012 1013
	spin_unlock_irqrestore(&log->io_list_lock, flags);
}

S
Shaohua Li 已提交
1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029
/*
 * Starting dispatch IO to raid.
 * io_unit(meta) consists of a log. There is one situation we want to avoid. A
 * broken meta in the middle of a log causes recovery can't find meta at the
 * head of log. If operations require meta at the head persistent in log, we
 * must make sure meta before it persistent in log too. A case is:
 *
 * stripe data/parity is in log, we start write stripe to raid disks. stripe
 * data/parity must be persistent in log before we do the write to raid disks.
 *
 * The solution is we restrictly maintain io_unit list order. In this case, we
 * only write stripes of an io_unit to raid disks till the io_unit is the first
 * one whose data/parity is in log.
 */
void r5l_flush_stripe_to_raid(struct r5l_log *log)
{
1030
	bool do_flush;
1031 1032

	if (!log || !log->need_cache_flush)
S
Shaohua Li 已提交
1033 1034 1035
		return;

	spin_lock_irq(&log->io_list_lock);
1036 1037 1038 1039
	/* flush bio is running */
	if (!list_empty(&log->flushing_ios)) {
		spin_unlock_irq(&log->io_list_lock);
		return;
S
Shaohua Li 已提交
1040
	}
1041 1042
	list_splice_tail_init(&log->io_end_ios, &log->flushing_ios);
	do_flush = !list_empty(&log->flushing_ios);
S
Shaohua Li 已提交
1043
	spin_unlock_irq(&log->io_list_lock);
1044 1045 1046 1047 1048 1049

	if (!do_flush)
		return;
	bio_reset(&log->flush_bio);
	log->flush_bio.bi_bdev = log->rdev->bdev;
	log->flush_bio.bi_end_io = r5l_log_flush_endio;
M
Mike Christie 已提交
1050
	bio_set_op_attrs(&log->flush_bio, REQ_OP_WRITE, WRITE_FLUSH);
1051
	submit_bio(&log->flush_bio);
S
Shaohua Li 已提交
1052 1053 1054
}

static void r5l_write_super(struct r5l_log *log, sector_t cp);
S
Shaohua Li 已提交
1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
static void r5l_write_super_and_discard_space(struct r5l_log *log,
	sector_t end)
{
	struct block_device *bdev = log->rdev->bdev;
	struct mddev *mddev;

	r5l_write_super(log, end);

	if (!blk_queue_discard(bdev_get_queue(bdev)))
		return;

	mddev = log->rdev->mddev;
	/*
1068 1069 1070 1071 1072 1073 1074 1075 1076
	 * Discard could zero data, so before discard we must make sure
	 * superblock is updated to new log tail. Updating superblock (either
	 * directly call md_update_sb() or depend on md thread) must hold
	 * reconfig mutex. On the other hand, raid5_quiesce is called with
	 * reconfig_mutex hold. The first step of raid5_quiesce() is waitting
	 * for all IO finish, hence waitting for reclaim thread, while reclaim
	 * thread is calling this function and waitting for reconfig mutex. So
	 * there is a deadlock. We workaround this issue with a trylock.
	 * FIXME: we could miss discard if we can't take reconfig mutex
S
Shaohua Li 已提交
1077
	 */
1078 1079 1080 1081 1082 1083
	set_mask_bits(&mddev->flags, 0,
		BIT(MD_CHANGE_DEVS) | BIT(MD_CHANGE_PENDING));
	if (!mddev_trylock(mddev))
		return;
	md_update_sb(mddev, 1);
	mddev_unlock(mddev);
S
Shaohua Li 已提交
1084

S
Shaohua Li 已提交
1085
	/* discard IO error really doesn't matter, ignore it */
S
Shaohua Li 已提交
1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
	if (log->last_checkpoint < end) {
		blkdev_issue_discard(bdev,
				log->last_checkpoint + log->rdev->data_offset,
				end - log->last_checkpoint, GFP_NOIO, 0);
	} else {
		blkdev_issue_discard(bdev,
				log->last_checkpoint + log->rdev->data_offset,
				log->device_size - log->last_checkpoint,
				GFP_NOIO, 0);
		blkdev_issue_discard(bdev, log->rdev->data_offset, end,
				GFP_NOIO, 0);
	}
}

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 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228
/*
 * r5c_flush_stripe moves stripe from cached list to handle_list. When called,
 * the stripe must be on r5c_cached_full_stripes or r5c_cached_partial_stripes.
 *
 * must hold conf->device_lock
 */
static void r5c_flush_stripe(struct r5conf *conf, struct stripe_head *sh)
{
	BUG_ON(list_empty(&sh->lru));
	BUG_ON(!test_bit(STRIPE_R5C_CACHING, &sh->state));
	BUG_ON(test_bit(STRIPE_HANDLE, &sh->state));

	/*
	 * The stripe is not ON_RELEASE_LIST, so it is safe to call
	 * raid5_release_stripe() while holding conf->device_lock
	 */
	BUG_ON(test_bit(STRIPE_ON_RELEASE_LIST, &sh->state));
	assert_spin_locked(&conf->device_lock);

	list_del_init(&sh->lru);
	atomic_inc(&sh->count);

	set_bit(STRIPE_HANDLE, &sh->state);
	atomic_inc(&conf->active_stripes);
	r5c_make_stripe_write_out(sh);

	if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
		atomic_inc(&conf->preread_active_stripes);
	raid5_release_stripe(sh);
}

/*
 * if num == 0, flush all full stripes
 * if num > 0, flush all full stripes. If less than num full stripes are
 *             flushed, flush some partial stripes until totally num stripes are
 *             flushed or there is no more cached stripes.
 */
void r5c_flush_cache(struct r5conf *conf, int num)
{
	int count;
	struct stripe_head *sh, *next;

	assert_spin_locked(&conf->device_lock);
	if (!conf->log)
		return;

	count = 0;
	list_for_each_entry_safe(sh, next, &conf->r5c_full_stripe_list, lru) {
		r5c_flush_stripe(conf, sh);
		count++;
	}

	if (count >= num)
		return;
	list_for_each_entry_safe(sh, next,
				 &conf->r5c_partial_stripe_list, lru) {
		r5c_flush_stripe(conf, sh);
		if (++count >= num)
			break;
	}
}

static void r5c_do_reclaim(struct r5conf *conf)
{
	struct r5l_log *log = conf->log;
	struct stripe_head *sh;
	int count = 0;
	unsigned long flags;
	int total_cached;
	int stripes_to_flush;

	if (!r5c_is_writeback(log))
		return;

	total_cached = atomic_read(&conf->r5c_cached_partial_stripes) +
		atomic_read(&conf->r5c_cached_full_stripes);

	if (total_cached > conf->min_nr_stripes * 3 / 4 ||
	    atomic_read(&conf->empty_inactive_list_nr) > 0)
		/*
		 * if stripe cache pressure high, flush all full stripes and
		 * some partial stripes
		 */
		stripes_to_flush = R5C_RECLAIM_STRIPE_GROUP;
	else if (total_cached > conf->min_nr_stripes * 1 / 2 ||
		 atomic_read(&conf->r5c_cached_full_stripes) >
		 R5C_FULL_STRIPE_FLUSH_BATCH)
		/*
		 * if stripe cache pressure moderate, or if there is many full
		 * stripes,flush all full stripes
		 */
		stripes_to_flush = 0;
	else
		/* no need to flush */
		stripes_to_flush = -1;

	if (stripes_to_flush >= 0) {
		spin_lock_irqsave(&conf->device_lock, flags);
		r5c_flush_cache(conf, stripes_to_flush);
		spin_unlock_irqrestore(&conf->device_lock, flags);
	}

	/* if log space is tight, flush stripes on stripe_in_journal_list */
	if (test_bit(R5C_LOG_TIGHT, &conf->cache_state)) {
		spin_lock_irqsave(&log->stripe_in_journal_lock, flags);
		spin_lock(&conf->device_lock);
		list_for_each_entry(sh, &log->stripe_in_journal_list, r5c) {
			/*
			 * stripes on stripe_in_journal_list could be in any
			 * state of the stripe_cache state machine. In this
			 * case, we only want to flush stripe on
			 * r5c_cached_full/partial_stripes. The following
			 * condition makes sure the stripe is on one of the
			 * two lists.
			 */
			if (!list_empty(&sh->lru) &&
			    !test_bit(STRIPE_HANDLE, &sh->state) &&
			    atomic_read(&sh->count) == 0) {
				r5c_flush_stripe(conf, sh);
			}
			if (count++ >= R5C_RECLAIM_STRIPE_GROUP)
				break;
		}
		spin_unlock(&conf->device_lock);
		spin_unlock_irqrestore(&log->stripe_in_journal_lock, flags);
	}
	md_wakeup_thread(conf->mddev->thread);
}

S
Shaohua Li 已提交
1229 1230
static void r5l_do_reclaim(struct r5l_log *log)
{
1231
	struct r5conf *conf = log->rdev->mddev->private;
S
Shaohua Li 已提交
1232
	sector_t reclaim_target = xchg(&log->reclaim_target, 0);
1233 1234
	sector_t reclaimable;
	sector_t next_checkpoint;
1235
	bool write_super;
S
Shaohua Li 已提交
1236 1237

	spin_lock_irq(&log->io_list_lock);
1238 1239
	write_super = r5l_reclaimable_space(log) > log->max_free_space ||
		reclaim_target != 0 || !list_empty(&log->no_space_stripes);
S
Shaohua Li 已提交
1240 1241 1242 1243 1244 1245
	/*
	 * move proper io_unit to reclaim list. We should not change the order.
	 * reclaimable/unreclaimable io_unit can be mixed in the list, we
	 * shouldn't reuse space of an unreclaimable io_unit
	 */
	while (1) {
1246 1247
		reclaimable = r5l_reclaimable_space(log);
		if (reclaimable >= reclaim_target ||
S
Shaohua Li 已提交
1248 1249
		    (list_empty(&log->running_ios) &&
		     list_empty(&log->io_end_ios) &&
1250
		     list_empty(&log->flushing_ios) &&
1251
		     list_empty(&log->finished_ios)))
S
Shaohua Li 已提交
1252 1253
			break;

1254 1255 1256 1257
		md_wakeup_thread(log->rdev->mddev->thread);
		wait_event_lock_irq(log->iounit_wait,
				    r5l_reclaimable_space(log) > reclaimable,
				    log->io_list_lock);
S
Shaohua Li 已提交
1258
	}
1259

1260
	next_checkpoint = r5c_calculate_new_cp(conf);
S
Shaohua Li 已提交
1261 1262
	spin_unlock_irq(&log->io_list_lock);

1263
	BUG_ON(reclaimable < 0);
1264 1265

	if (reclaimable == 0 || !write_super)
S
Shaohua Li 已提交
1266 1267 1268 1269 1270 1271 1272
		return;

	/*
	 * write_super will flush cache of each raid disk. We must write super
	 * here, because the log area might be reused soon and we don't want to
	 * confuse recovery
	 */
S
Shaohua Li 已提交
1273
	r5l_write_super_and_discard_space(log, next_checkpoint);
S
Shaohua Li 已提交
1274 1275

	mutex_lock(&log->io_mutex);
1276
	log->last_checkpoint = next_checkpoint;
1277
	r5c_update_log_state(log);
S
Shaohua Li 已提交
1278 1279
	mutex_unlock(&log->io_mutex);

1280
	r5l_run_no_space_stripes(log);
S
Shaohua Li 已提交
1281 1282 1283 1284 1285 1286 1287 1288 1289 1290
}

static void r5l_reclaim_thread(struct md_thread *thread)
{
	struct mddev *mddev = thread->mddev;
	struct r5conf *conf = mddev->private;
	struct r5l_log *log = conf->log;

	if (!log)
		return;
1291
	r5c_do_reclaim(conf);
S
Shaohua Li 已提交
1292 1293 1294
	r5l_do_reclaim(log);
}

1295
void r5l_wake_reclaim(struct r5l_log *log, sector_t space)
S
Shaohua Li 已提交
1296
{
S
Shaohua Li 已提交
1297 1298 1299
	unsigned long target;
	unsigned long new = (unsigned long)space; /* overflow in theory */

1300 1301
	if (!log)
		return;
S
Shaohua Li 已提交
1302 1303 1304 1305 1306 1307
	do {
		target = log->reclaim_target;
		if (new < target)
			return;
	} while (cmpxchg(&log->reclaim_target, target, new) != target);
	md_wakeup_thread(log->reclaim_thread);
S
Shaohua Li 已提交
1308 1309
}

1310 1311
void r5l_quiesce(struct r5l_log *log, int state)
{
S
Shaohua Li 已提交
1312
	struct mddev *mddev;
1313 1314 1315
	if (!log || state == 2)
		return;
	if (state == 0) {
1316 1317 1318 1319 1320 1321 1322
		/*
		 * This is a special case for hotadd. In suspend, the array has
		 * no journal. In resume, journal is initialized as well as the
		 * reclaim thread.
		 */
		if (log->reclaim_thread)
			return;
1323 1324
		log->reclaim_thread = md_register_thread(r5l_reclaim_thread,
					log->rdev->mddev, "reclaim");
1325
		log->reclaim_thread->timeout = R5C_RECLAIM_WAKEUP_INTERVAL;
1326
	} else if (state == 1) {
S
Shaohua Li 已提交
1327 1328 1329
		/* make sure r5l_write_super_and_discard_space exits */
		mddev = log->rdev->mddev;
		wake_up(&mddev->sb_wait);
1330
		r5l_wake_reclaim(log, MaxSector);
1331 1332 1333 1334 1335
		md_unregister_thread(&log->reclaim_thread);
		r5l_do_reclaim(log);
	}
}

S
Shaohua Li 已提交
1336 1337
bool r5l_log_disk_error(struct r5conf *conf)
{
1338 1339
	struct r5l_log *log;
	bool ret;
1340
	/* don't allow write if journal disk is missing */
1341 1342 1343 1344 1345 1346 1347 1348 1349
	rcu_read_lock();
	log = rcu_dereference(conf->log);

	if (!log)
		ret = test_bit(MD_HAS_JOURNAL, &conf->mddev->flags);
	else
		ret = test_bit(Faulty, &log->rdev->flags);
	rcu_read_unlock();
	return ret;
S
Shaohua Li 已提交
1350 1351
}

S
Shaohua Li 已提交
1352 1353 1354 1355 1356 1357 1358
struct r5l_recovery_ctx {
	struct page *meta_page;		/* current meta */
	sector_t meta_total_blocks;	/* total size of current meta and data */
	sector_t pos;			/* recovery position */
	u64 seq;			/* recovery position seq */
};

1359 1360
static int r5l_recovery_read_meta_block(struct r5l_log *log,
					struct r5l_recovery_ctx *ctx)
S
Shaohua Li 已提交
1361 1362 1363 1364 1365
{
	struct page *page = ctx->meta_page;
	struct r5l_meta_block *mb;
	u32 crc, stored_crc;

M
Mike Christie 已提交
1366 1367
	if (!sync_page_io(log->rdev, ctx->pos, PAGE_SIZE, page, REQ_OP_READ, 0,
			  false))
S
Shaohua Li 已提交
1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379
		return -EIO;

	mb = page_address(page);
	stored_crc = le32_to_cpu(mb->checksum);
	mb->checksum = 0;

	if (le32_to_cpu(mb->magic) != R5LOG_MAGIC ||
	    le64_to_cpu(mb->seq) != ctx->seq ||
	    mb->version != R5LOG_VERSION ||
	    le64_to_cpu(mb->position) != ctx->pos)
		return -EINVAL;

S
Shaohua Li 已提交
1380
	crc = crc32c_le(log->uuid_checksum, mb, PAGE_SIZE);
S
Shaohua Li 已提交
1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394
	if (stored_crc != crc)
		return -EINVAL;

	if (le32_to_cpu(mb->meta_size) > PAGE_SIZE)
		return -EINVAL;

	ctx->meta_total_blocks = BLOCK_SECTORS;

	return 0;
}

static int r5l_recovery_flush_one_stripe(struct r5l_log *log,
					 struct r5l_recovery_ctx *ctx,
					 sector_t stripe_sect,
1395
					 int *offset)
S
Shaohua Li 已提交
1396 1397 1398 1399 1400 1401 1402 1403
{
	struct r5conf *conf = log->rdev->mddev->private;
	struct stripe_head *sh;
	struct r5l_payload_data_parity *payload;
	int disk_index;

	sh = raid5_get_active_stripe(conf, stripe_sect, 0, 0, 0);
	while (1) {
1404 1405
		sector_t log_offset = r5l_ring_add(log, ctx->pos,
				ctx->meta_total_blocks);
S
Shaohua Li 已提交
1406 1407 1408 1409 1410 1411 1412
		payload = page_address(ctx->meta_page) + *offset;

		if (le16_to_cpu(payload->header.type) == R5LOG_PAYLOAD_DATA) {
			raid5_compute_sector(conf,
					     le64_to_cpu(payload->location), 0,
					     &disk_index, sh);

1413
			sync_page_io(log->rdev, log_offset, PAGE_SIZE,
M
Mike Christie 已提交
1414 1415
				     sh->dev[disk_index].page, REQ_OP_READ, 0,
				     false);
S
Shaohua Li 已提交
1416 1417 1418 1419 1420
			sh->dev[disk_index].log_checksum =
				le32_to_cpu(payload->checksum[0]);
			set_bit(R5_Wantwrite, &sh->dev[disk_index].flags);
		} else {
			disk_index = sh->pd_idx;
1421
			sync_page_io(log->rdev, log_offset, PAGE_SIZE,
M
Mike Christie 已提交
1422 1423
				     sh->dev[disk_index].page, REQ_OP_READ, 0,
				     false);
S
Shaohua Li 已提交
1424 1425 1426 1427 1428 1429 1430
			sh->dev[disk_index].log_checksum =
				le32_to_cpu(payload->checksum[0]);
			set_bit(R5_Wantwrite, &sh->dev[disk_index].flags);

			if (sh->qd_idx >= 0) {
				disk_index = sh->qd_idx;
				sync_page_io(log->rdev,
1431
					     r5l_ring_add(log, log_offset, BLOCK_SECTORS),
S
Shaohua Li 已提交
1432
					     PAGE_SIZE, sh->dev[disk_index].page,
M
Mike Christie 已提交
1433
					     REQ_OP_READ, 0, false);
S
Shaohua Li 已提交
1434 1435 1436 1437 1438 1439 1440
				sh->dev[disk_index].log_checksum =
					le32_to_cpu(payload->checksum[1]);
				set_bit(R5_Wantwrite,
					&sh->dev[disk_index].flags);
			}
		}

1441
		ctx->meta_total_blocks += le32_to_cpu(payload->size);
S
Shaohua Li 已提交
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
		*offset += sizeof(struct r5l_payload_data_parity) +
			sizeof(__le32) *
			(le32_to_cpu(payload->size) >> (PAGE_SHIFT - 9));
		if (le16_to_cpu(payload->header.type) == R5LOG_PAYLOAD_PARITY)
			break;
	}

	for (disk_index = 0; disk_index < sh->disks; disk_index++) {
		void *addr;
		u32 checksum;

		if (!test_bit(R5_Wantwrite, &sh->dev[disk_index].flags))
			continue;
		addr = kmap_atomic(sh->dev[disk_index].page);
S
Shaohua Li 已提交
1456
		checksum = crc32c_le(log->uuid_checksum, addr, PAGE_SIZE);
S
Shaohua Li 已提交
1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469
		kunmap_atomic(addr);
		if (checksum != sh->dev[disk_index].log_checksum)
			goto error;
	}

	for (disk_index = 0; disk_index < sh->disks; disk_index++) {
		struct md_rdev *rdev, *rrdev;

		if (!test_and_clear_bit(R5_Wantwrite,
					&sh->dev[disk_index].flags))
			continue;

		/* in case device is broken */
S
Shaohua Li 已提交
1470
		rcu_read_lock();
S
Shaohua Li 已提交
1471
		rdev = rcu_dereference(conf->disks[disk_index].rdev);
S
Shaohua Li 已提交
1472 1473 1474
		if (rdev) {
			atomic_inc(&rdev->nr_pending);
			rcu_read_unlock();
S
Shaohua Li 已提交
1475
			sync_page_io(rdev, stripe_sect, PAGE_SIZE,
M
Mike Christie 已提交
1476 1477
				     sh->dev[disk_index].page, REQ_OP_WRITE, 0,
				     false);
S
Shaohua Li 已提交
1478 1479 1480
			rdev_dec_pending(rdev, rdev->mddev);
			rcu_read_lock();
		}
S
Shaohua Li 已提交
1481
		rrdev = rcu_dereference(conf->disks[disk_index].replacement);
S
Shaohua Li 已提交
1482 1483 1484
		if (rrdev) {
			atomic_inc(&rrdev->nr_pending);
			rcu_read_unlock();
S
Shaohua Li 已提交
1485
			sync_page_io(rrdev, stripe_sect, PAGE_SIZE,
M
Mike Christie 已提交
1486 1487
				     sh->dev[disk_index].page, REQ_OP_WRITE, 0,
				     false);
S
Shaohua Li 已提交
1488 1489 1490 1491
			rdev_dec_pending(rrdev, rrdev->mddev);
			rcu_read_lock();
		}
		rcu_read_unlock();
S
Shaohua Li 已提交
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
	}
	raid5_release_stripe(sh);
	return 0;

error:
	for (disk_index = 0; disk_index < sh->disks; disk_index++)
		sh->dev[disk_index].flags = 0;
	raid5_release_stripe(sh);
	return -EINVAL;
}

static int r5l_recovery_flush_one_meta(struct r5l_log *log,
				       struct r5l_recovery_ctx *ctx)
{
	struct r5conf *conf = log->rdev->mddev->private;
	struct r5l_payload_data_parity *payload;
	struct r5l_meta_block *mb;
	int offset;
	sector_t stripe_sector;

	mb = page_address(ctx->meta_page);
	offset = sizeof(struct r5l_meta_block);

	while (offset < le32_to_cpu(mb->meta_size)) {
		int dd;

		payload = (void *)mb + offset;
		stripe_sector = raid5_compute_sector(conf,
						     le64_to_cpu(payload->location), 0, &dd, NULL);
		if (r5l_recovery_flush_one_stripe(log, ctx, stripe_sector,
1522
						  &offset))
S
Shaohua Li 已提交
1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
			return -EINVAL;
	}
	return 0;
}

/* copy data/parity from log to raid disks */
static void r5l_recovery_flush_log(struct r5l_log *log,
				   struct r5l_recovery_ctx *ctx)
{
	while (1) {
1533
		if (r5l_recovery_read_meta_block(log, ctx))
S
Shaohua Li 已提交
1534 1535 1536 1537 1538 1539 1540 1541
			return;
		if (r5l_recovery_flush_one_meta(log, ctx))
			return;
		ctx->seq++;
		ctx->pos = r5l_ring_add(log, ctx->pos, ctx->meta_total_blocks);
	}
}

1542 1543 1544 1545
static void
r5l_recovery_create_empty_meta_block(struct r5l_log *log,
				     struct page *page,
				     sector_t pos, u64 seq)
S
Shaohua Li 已提交
1546 1547 1548 1549 1550
{
	struct r5l_meta_block *mb;
	u32 crc;

	mb = page_address(page);
1551
	clear_page(mb);
S
Shaohua Li 已提交
1552 1553 1554 1555 1556
	mb->magic = cpu_to_le32(R5LOG_MAGIC);
	mb->version = R5LOG_VERSION;
	mb->meta_size = cpu_to_le32(sizeof(struct r5l_meta_block));
	mb->seq = cpu_to_le64(seq);
	mb->position = cpu_to_le64(pos);
S
Shaohua Li 已提交
1557
	crc = crc32c_le(log->uuid_checksum, mb, PAGE_SIZE);
S
Shaohua Li 已提交
1558
	mb->checksum = cpu_to_le32(crc);
1559
}
S
Shaohua Li 已提交
1560

1561 1562 1563 1564 1565 1566 1567 1568 1569
static int r5l_log_write_empty_meta_block(struct r5l_log *log, sector_t pos,
					  u64 seq)
{
	struct page *page;

	page = alloc_page(GFP_KERNEL);
	if (!page)
		return -ENOMEM;
	r5l_recovery_create_empty_meta_block(log, page, pos, seq);
M
Mike Christie 已提交
1570 1571
	if (!sync_page_io(log->rdev, pos, PAGE_SIZE, page, REQ_OP_WRITE,
			  WRITE_FUA, false)) {
S
Shaohua Li 已提交
1572 1573 1574 1575 1576 1577 1578
		__free_page(page);
		return -EIO;
	}
	__free_page(page);
	return 0;
}

S
Shaohua Li 已提交
1579 1580
static int r5l_recovery_log(struct r5l_log *log)
{
S
Shaohua Li 已提交
1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604
	struct r5l_recovery_ctx ctx;

	ctx.pos = log->last_checkpoint;
	ctx.seq = log->last_cp_seq;
	ctx.meta_page = alloc_page(GFP_KERNEL);
	if (!ctx.meta_page)
		return -ENOMEM;

	r5l_recovery_flush_log(log, &ctx);
	__free_page(ctx.meta_page);

	/*
	 * we did a recovery. Now ctx.pos points to an invalid meta block. New
	 * log will start here. but we can't let superblock point to last valid
	 * meta block. The log might looks like:
	 * | meta 1| meta 2| meta 3|
	 * meta 1 is valid, meta 2 is invalid. meta 3 could be valid. If
	 * superblock points to meta 1, we write a new valid meta 2n.  if crash
	 * happens again, new recovery will start from meta 1. Since meta 2n is
	 * valid now, recovery will think meta 3 is valid, which is wrong.
	 * The solution is we create a new meta in meta2 with its seq == meta
	 * 1's seq + 10 and let superblock points to meta2. The same recovery will
	 * not think meta 3 is a valid meta, because its seq doesn't match
	 */
1605
	if (ctx.seq > log->last_cp_seq) {
S
Shaohua Li 已提交
1606 1607 1608 1609 1610 1611 1612 1613
		int ret;

		ret = r5l_log_write_empty_meta_block(log, ctx.pos, ctx.seq + 10);
		if (ret)
			return ret;
		log->seq = ctx.seq + 11;
		log->log_start = r5l_ring_add(log, ctx.pos, BLOCK_SECTORS);
		r5l_write_super(log, ctx.pos);
1614 1615
		log->last_checkpoint = ctx.pos;
		log->next_checkpoint = ctx.pos;
S
Shaohua Li 已提交
1616 1617 1618 1619
	} else {
		log->log_start = ctx.pos;
		log->seq = ctx.seq;
	}
S
Shaohua Li 已提交
1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630
	return 0;
}

static void r5l_write_super(struct r5l_log *log, sector_t cp)
{
	struct mddev *mddev = log->rdev->mddev;

	log->rdev->journal_tail = cp;
	set_bit(MD_CHANGE_DEVS, &mddev->flags);
}

S
Song Liu 已提交
1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693
static ssize_t r5c_journal_mode_show(struct mddev *mddev, char *page)
{
	struct r5conf *conf = mddev->private;
	int ret;

	if (!conf->log)
		return 0;

	switch (conf->log->r5c_journal_mode) {
	case R5C_JOURNAL_MODE_WRITE_THROUGH:
		ret = snprintf(
			page, PAGE_SIZE, "[%s] %s\n",
			r5c_journal_mode_str[R5C_JOURNAL_MODE_WRITE_THROUGH],
			r5c_journal_mode_str[R5C_JOURNAL_MODE_WRITE_BACK]);
		break;
	case R5C_JOURNAL_MODE_WRITE_BACK:
		ret = snprintf(
			page, PAGE_SIZE, "%s [%s]\n",
			r5c_journal_mode_str[R5C_JOURNAL_MODE_WRITE_THROUGH],
			r5c_journal_mode_str[R5C_JOURNAL_MODE_WRITE_BACK]);
		break;
	default:
		ret = 0;
	}
	return ret;
}

static ssize_t r5c_journal_mode_store(struct mddev *mddev,
				      const char *page, size_t length)
{
	struct r5conf *conf = mddev->private;
	struct r5l_log *log = conf->log;
	int val = -1, i;
	int len = length;

	if (!log)
		return -ENODEV;

	if (len && page[len - 1] == '\n')
		len -= 1;
	for (i = 0; i < ARRAY_SIZE(r5c_journal_mode_str); i++)
		if (strlen(r5c_journal_mode_str[i]) == len &&
		    strncmp(page, r5c_journal_mode_str[i], len) == 0) {
			val = i;
			break;
		}
	if (val < R5C_JOURNAL_MODE_WRITE_THROUGH ||
	    val > R5C_JOURNAL_MODE_WRITE_BACK)
		return -EINVAL;

	mddev_suspend(mddev);
	conf->log->r5c_journal_mode = val;
	mddev_resume(mddev);

	pr_debug("md/raid:%s: setting r5c cache mode to %d: %s\n",
		 mdname(mddev), val, r5c_journal_mode_str[val]);
	return length;
}

struct md_sysfs_entry
r5c_journal_mode = __ATTR(journal_mode, 0644,
			  r5c_journal_mode_show, r5c_journal_mode_store);

1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707
/*
 * Try handle write operation in caching phase. This function should only
 * be called in write-back mode.
 *
 * If all outstanding writes can be handled in caching phase, returns 0
 * If writes requires write-out phase, call r5c_make_stripe_write_out()
 * and returns -EAGAIN
 */
int r5c_try_caching_write(struct r5conf *conf,
			  struct stripe_head *sh,
			  struct stripe_head_state *s,
			  int disks)
{
	struct r5l_log *log = conf->log;
S
Song Liu 已提交
1708 1709 1710
	int i;
	struct r5dev *dev;
	int to_cache = 0;
1711 1712 1713

	BUG_ON(!r5c_is_writeback(log));

S
Song Liu 已提交
1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781
	if (!test_bit(STRIPE_R5C_CACHING, &sh->state)) {
		/*
		 * There are two different scenarios here:
		 *  1. The stripe has some data cached, and it is sent to
		 *     write-out phase for reclaim
		 *  2. The stripe is clean, and this is the first write
		 *
		 * For 1, return -EAGAIN, so we continue with
		 * handle_stripe_dirtying().
		 *
		 * For 2, set STRIPE_R5C_CACHING and continue with caching
		 * write.
		 */

		/* case 1: anything injournal or anything in written */
		if (s->injournal > 0 || s->written > 0)
			return -EAGAIN;
		/* case 2 */
		set_bit(STRIPE_R5C_CACHING, &sh->state);
	}

	for (i = disks; i--; ) {
		dev = &sh->dev[i];
		/* if non-overwrite, use writing-out phase */
		if (dev->towrite && !test_bit(R5_OVERWRITE, &dev->flags) &&
		    !test_bit(R5_InJournal, &dev->flags)) {
			r5c_make_stripe_write_out(sh);
			return -EAGAIN;
		}
	}

	for (i = disks; i--; ) {
		dev = &sh->dev[i];
		if (dev->towrite) {
			set_bit(R5_Wantwrite, &dev->flags);
			set_bit(R5_Wantdrain, &dev->flags);
			set_bit(R5_LOCKED, &dev->flags);
			to_cache++;
		}
	}

	if (to_cache) {
		set_bit(STRIPE_OP_BIODRAIN, &s->ops_request);
		/*
		 * set STRIPE_LOG_TRAPPED, which triggers r5c_cache_data()
		 * in ops_run_io(). STRIPE_LOG_TRAPPED will be cleared in
		 * r5c_handle_data_cached()
		 */
		set_bit(STRIPE_LOG_TRAPPED, &sh->state);
	}

	return 0;
}

/*
 * free extra pages (orig_page) we allocated for prexor
 */
void r5c_release_extra_page(struct stripe_head *sh)
{
	int i;

	for (i = sh->disks; i--; )
		if (sh->dev[i].page != sh->dev[i].orig_page) {
			struct page *p = sh->dev[i].orig_page;

			sh->dev[i].orig_page = sh->dev[i].page;
			put_page(p);
		}
1782 1783 1784 1785 1786 1787 1788 1789 1790 1791
}

/*
 * clean up the stripe (clear R5_InJournal for dev[pd_idx] etc.) after the
 * stripe is committed to RAID disks.
 */
void r5c_finish_stripe_write_out(struct r5conf *conf,
				 struct stripe_head *sh,
				 struct stripe_head_state *s)
{
S
Song Liu 已提交
1792 1793 1794
	int i;
	int do_wakeup = 0;

1795 1796 1797 1798 1799 1800 1801 1802 1803
	if (!conf->log ||
	    !test_bit(R5_InJournal, &sh->dev[sh->pd_idx].flags))
		return;

	WARN_ON(test_bit(STRIPE_R5C_CACHING, &sh->state));
	clear_bit(R5_InJournal, &sh->dev[sh->pd_idx].flags);

	if (conf->log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH)
		return;
S
Song Liu 已提交
1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822

	for (i = sh->disks; i--; ) {
		clear_bit(R5_InJournal, &sh->dev[i].flags);
		if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
			do_wakeup = 1;
	}

	/*
	 * analyse_stripe() runs before r5c_finish_stripe_write_out(),
	 * We updated R5_InJournal, so we also update s->injournal.
	 */
	s->injournal = 0;

	if (test_and_clear_bit(STRIPE_FULL_WRITE, &sh->state))
		if (atomic_dec_and_test(&conf->pending_full_writes))
			md_wakeup_thread(conf->mddev->thread);

	if (do_wakeup)
		wake_up(&conf->wait_for_overlap);
1823 1824 1825 1826 1827 1828 1829 1830 1831

	if (conf->log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH)
		return;

	spin_lock_irq(&conf->log->stripe_in_journal_lock);
	list_del_init(&sh->r5c);
	spin_unlock_irq(&conf->log->stripe_in_journal_lock);
	sh->log_start = MaxSector;
	atomic_dec(&conf->log->stripe_in_journal_count);
S
Song Liu 已提交
1832 1833 1834 1835 1836 1837
}

int
r5c_cache_data(struct r5l_log *log, struct stripe_head *sh,
	       struct stripe_head_state *s)
{
1838
	struct r5conf *conf = sh->raid_conf;
S
Song Liu 已提交
1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869
	int pages = 0;
	int reserve;
	int i;
	int ret = 0;

	BUG_ON(!log);

	for (i = 0; i < sh->disks; i++) {
		void *addr;

		if (!test_bit(R5_Wantwrite, &sh->dev[i].flags))
			continue;
		addr = kmap_atomic(sh->dev[i].page);
		sh->dev[i].log_checksum = crc32c_le(log->uuid_checksum,
						    addr, PAGE_SIZE);
		kunmap_atomic(addr);
		pages++;
	}
	WARN_ON(pages == 0);

	/*
	 * The stripe must enter state machine again to call endio, so
	 * don't delay.
	 */
	clear_bit(STRIPE_DELAYED, &sh->state);
	atomic_inc(&sh->count);

	mutex_lock(&log->io_mutex);
	/* meta + data */
	reserve = (1 + pages) << (PAGE_SHIFT - 9);

1870 1871 1872 1873 1874 1875 1876 1877
	if (test_bit(R5C_LOG_CRITICAL, &conf->cache_state) &&
	    sh->log_start == MaxSector)
		r5l_add_no_space_stripe(log, sh);
	else if (!r5l_has_free_space(log, reserve)) {
		if (sh->log_start == log->last_checkpoint)
			BUG();
		else
			r5l_add_no_space_stripe(log, sh);
S
Song Liu 已提交
1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888
	} else {
		ret = r5l_log_stripe(log, sh, pages, 0);
		if (ret) {
			spin_lock_irq(&log->io_list_lock);
			list_add_tail(&sh->log_list, &log->no_mem_stripes);
			spin_unlock_irq(&log->io_list_lock);
		}
	}

	mutex_unlock(&log->io_mutex);
	return 0;
1889 1890
}

S
Shaohua Li 已提交
1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907
static int r5l_load_log(struct r5l_log *log)
{
	struct md_rdev *rdev = log->rdev;
	struct page *page;
	struct r5l_meta_block *mb;
	sector_t cp = log->rdev->journal_tail;
	u32 stored_crc, expected_crc;
	bool create_super = false;
	int ret;

	/* Make sure it's valid */
	if (cp >= rdev->sectors || round_down(cp, BLOCK_SECTORS) != cp)
		cp = 0;
	page = alloc_page(GFP_KERNEL);
	if (!page)
		return -ENOMEM;

M
Mike Christie 已提交
1908
	if (!sync_page_io(rdev, cp, PAGE_SIZE, page, REQ_OP_READ, 0, false)) {
S
Shaohua Li 已提交
1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920
		ret = -EIO;
		goto ioerr;
	}
	mb = page_address(page);

	if (le32_to_cpu(mb->magic) != R5LOG_MAGIC ||
	    mb->version != R5LOG_VERSION) {
		create_super = true;
		goto create;
	}
	stored_crc = le32_to_cpu(mb->checksum);
	mb->checksum = 0;
S
Shaohua Li 已提交
1921
	expected_crc = crc32c_le(log->uuid_checksum, mb, PAGE_SIZE);
S
Shaohua Li 已提交
1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933
	if (stored_crc != expected_crc) {
		create_super = true;
		goto create;
	}
	if (le64_to_cpu(mb->position) != cp) {
		create_super = true;
		goto create;
	}
create:
	if (create_super) {
		log->last_cp_seq = prandom_u32();
		cp = 0;
1934
		r5l_log_write_empty_meta_block(log, cp, log->last_cp_seq);
S
Shaohua Li 已提交
1935 1936 1937 1938 1939 1940 1941 1942 1943 1944
		/*
		 * Make sure super points to correct address. Log might have
		 * data very soon. If super hasn't correct log tail address,
		 * recovery can't find the log
		 */
		r5l_write_super(log, cp);
	} else
		log->last_cp_seq = le64_to_cpu(mb->seq);

	log->device_size = round_down(rdev->sectors, BLOCK_SECTORS);
S
Shaohua Li 已提交
1945 1946 1947
	log->max_free_space = log->device_size >> RECLAIM_MAX_FREE_SPACE_SHIFT;
	if (log->max_free_space > RECLAIM_MAX_FREE_SPACE)
		log->max_free_space = RECLAIM_MAX_FREE_SPACE;
S
Shaohua Li 已提交
1948
	log->last_checkpoint = cp;
1949
	log->next_checkpoint = cp;
1950 1951 1952
	mutex_lock(&log->io_mutex);
	r5c_update_log_state(log);
	mutex_unlock(&log->io_mutex);
S
Shaohua Li 已提交
1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963

	__free_page(page);

	return r5l_recovery_log(log);
ioerr:
	__free_page(page);
	return ret;
}

int r5l_init_log(struct r5conf *conf, struct md_rdev *rdev)
{
J
Jens Axboe 已提交
1964
	struct request_queue *q = bdev_get_queue(rdev->bdev);
S
Shaohua Li 已提交
1965 1966 1967 1968
	struct r5l_log *log;

	if (PAGE_SIZE != 4096)
		return -EINVAL;
1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984

	/*
	 * The PAGE_SIZE must be big enough to hold 1 r5l_meta_block and
	 * raid_disks r5l_payload_data_parity.
	 *
	 * Write journal and cache does not work for very big array
	 * (raid_disks > 203)
	 */
	if (sizeof(struct r5l_meta_block) +
	    ((sizeof(struct r5l_payload_data_parity) + sizeof(__le32)) *
	     conf->raid_disks) > PAGE_SIZE) {
		pr_err("md/raid:%s: write journal/cache doesn't work for array with %d disks\n",
		       mdname(conf->mddev), conf->raid_disks);
		return -EINVAL;
	}

S
Shaohua Li 已提交
1985 1986 1987 1988 1989
	log = kzalloc(sizeof(*log), GFP_KERNEL);
	if (!log)
		return -ENOMEM;
	log->rdev = rdev;

J
Jens Axboe 已提交
1990
	log->need_cache_flush = test_bit(QUEUE_FLAG_WC, &q->queue_flags) != 0;
1991

S
Shaohua Li 已提交
1992 1993
	log->uuid_checksum = crc32c_le(~0, rdev->mddev->uuid,
				       sizeof(rdev->mddev->uuid));
S
Shaohua Li 已提交
1994 1995 1996 1997 1998

	mutex_init(&log->io_mutex);

	spin_lock_init(&log->io_list_lock);
	INIT_LIST_HEAD(&log->running_ios);
S
Shaohua Li 已提交
1999
	INIT_LIST_HEAD(&log->io_end_ios);
2000
	INIT_LIST_HEAD(&log->flushing_ios);
2001
	INIT_LIST_HEAD(&log->finished_ios);
2002
	bio_init(&log->flush_bio);
S
Shaohua Li 已提交
2003 2004 2005 2006 2007

	log->io_kc = KMEM_CACHE(r5l_io_unit, 0);
	if (!log->io_kc)
		goto io_kc;

2008 2009 2010 2011
	log->io_pool = mempool_create_slab_pool(R5L_POOL_SIZE, log->io_kc);
	if (!log->io_pool)
		goto io_pool;

C
Christoph Hellwig 已提交
2012 2013 2014 2015
	log->bs = bioset_create(R5L_POOL_SIZE, 0);
	if (!log->bs)
		goto io_bs;

2016 2017 2018 2019
	log->meta_pool = mempool_create_page_pool(R5L_POOL_SIZE, 0);
	if (!log->meta_pool)
		goto out_mempool;

S
Shaohua Li 已提交
2020 2021 2022 2023
	log->reclaim_thread = md_register_thread(r5l_reclaim_thread,
						 log->rdev->mddev, "reclaim");
	if (!log->reclaim_thread)
		goto reclaim_thread;
2024 2025
	log->reclaim_thread->timeout = R5C_RECLAIM_WAKEUP_INTERVAL;

2026
	init_waitqueue_head(&log->iounit_wait);
S
Shaohua Li 已提交
2027

2028 2029
	INIT_LIST_HEAD(&log->no_mem_stripes);

S
Shaohua Li 已提交
2030 2031 2032
	INIT_LIST_HEAD(&log->no_space_stripes);
	spin_lock_init(&log->no_space_stripes_lock);

2033
	log->r5c_journal_mode = R5C_JOURNAL_MODE_WRITE_THROUGH;
2034 2035 2036
	INIT_LIST_HEAD(&log->stripe_in_journal_list);
	spin_lock_init(&log->stripe_in_journal_lock);
	atomic_set(&log->stripe_in_journal_count, 0);
2037

S
Shaohua Li 已提交
2038 2039 2040
	if (r5l_load_log(log))
		goto error;

2041
	rcu_assign_pointer(conf->log, log);
2042
	set_bit(MD_HAS_JOURNAL, &conf->mddev->flags);
S
Shaohua Li 已提交
2043
	return 0;
2044

S
Shaohua Li 已提交
2045
error:
S
Shaohua Li 已提交
2046 2047
	md_unregister_thread(&log->reclaim_thread);
reclaim_thread:
2048 2049
	mempool_destroy(log->meta_pool);
out_mempool:
C
Christoph Hellwig 已提交
2050 2051
	bioset_free(log->bs);
io_bs:
2052 2053
	mempool_destroy(log->io_pool);
io_pool:
S
Shaohua Li 已提交
2054 2055 2056 2057 2058 2059 2060 2061
	kmem_cache_destroy(log->io_kc);
io_kc:
	kfree(log);
	return -EINVAL;
}

void r5l_exit_log(struct r5l_log *log)
{
S
Shaohua Li 已提交
2062
	md_unregister_thread(&log->reclaim_thread);
2063
	mempool_destroy(log->meta_pool);
C
Christoph Hellwig 已提交
2064
	bioset_free(log->bs);
2065
	mempool_destroy(log->io_pool);
S
Shaohua Li 已提交
2066 2067 2068
	kmem_cache_destroy(log->io_kc);
	kfree(log);
}