raid5-cache.c 79.5 KB
Newer Older
S
Shaohua Li 已提交
1 2
/*
 * Copyright (C) 2015 Shaohua Li <shli@fb.com>
S
Song Liu 已提交
3
 * Copyright (C) 2016 Song Liu <songliubraving@fb.com>
S
Shaohua Li 已提交
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
 *
 * 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
20
#include <linux/crc32c.h>
S
Shaohua Li 已提交
21
#include <linux/random.h>
22
#include <linux/kthread.h>
23
#include <linux/types.h>
S
Shaohua Li 已提交
24 25
#include "md.h"
#include "raid5.h"
S
Song Liu 已提交
26
#include "bitmap.h"
S
Shaohua Li 已提交
27 28 29 30 31 32 33

/*
 * 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
34
/*
35 36 37 38
 * 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
39 40 41 42
 */
#define RECLAIM_MAX_FREE_SPACE (10 * 1024 * 1024 * 2) /* sector */
#define RECLAIM_MAX_FREE_SPACE_SHIFT (2)

43 44 45 46 47 48 49
/* 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
50 51 52 53 54 55
/*
 * 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

56 57 58 59 60 61 62 63 64 65
/*
 * 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
66 67
static char *r5c_journal_mode_str[] = {"write-through",
				       "write-back"};
68 69 70
/*
 * raid5 cache state machine
 *
71
 * With the RAID cache, each stripe works in two phases:
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98
 *	- 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
99 100 101 102 103 104 105
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
106 107
	sector_t max_free_space;	/* reclaim run if free space is at
					 * this size */
S
Shaohua Li 已提交
108 109 110 111 112 113 114 115

	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 */

116 117
	sector_t next_checkpoint;

S
Shaohua Li 已提交
118 119 120 121 122 123 124 125 126 127
	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 */
128 129
	struct list_head flushing_ios;	/* io_units which are waiting for log
					 * cache flush */
130
	struct list_head finished_ios;	/* io_units which settle down in log disk */
131
	struct bio flush_bio;
S
Shaohua Li 已提交
132

133 134
	struct list_head no_mem_stripes;   /* pending stripes, -ENOMEM */

S
Shaohua Li 已提交
135
	struct kmem_cache *io_kc;
136
	mempool_t *io_pool;
C
Christoph Hellwig 已提交
137
	struct bio_set *bs;
138
	mempool_t *meta_pool;
S
Shaohua Li 已提交
139

S
Shaohua Li 已提交
140 141 142 143 144 145 146 147
	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) */
148
	wait_queue_head_t iounit_wait;
S
Shaohua Li 已提交
149

S
Shaohua Li 已提交
150 151
	struct list_head no_space_stripes; /* pending stripes, log has no space */
	spinlock_t no_space_stripes_lock;
152 153

	bool need_cache_flush;
154 155 156

	/* for r5c_cache */
	enum r5c_journal_mode r5c_journal_mode;
157 158 159 160 161 162

	/* 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
Song Liu 已提交
163 164 165

	/* to submit async io_units, to fulfill ordering of flush */
	struct work_struct deferred_io_work;
166 167
	/* to disable write back during in degraded mode */
	struct work_struct disable_writeback_work;
168 169 170 171

	/* to for chunk_aligned_read in writeback mode, details below */
	spinlock_t tree_lock;
	struct radix_tree_root big_stripe_tree;
S
Shaohua Li 已提交
172 173
};

174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
/*
 * Enable chunk_aligned_read() with write back cache.
 *
 * Each chunk may contain more than one stripe (for example, a 256kB
 * chunk contains 64 4kB-page, so this chunk contain 64 stripes). For
 * chunk_aligned_read, these stripes are grouped into one "big_stripe".
 * For each big_stripe, we count how many stripes of this big_stripe
 * are in the write back cache. These data are tracked in a radix tree
 * (big_stripe_tree). We use radix_tree item pointer as the counter.
 * r5c_tree_index() is used to calculate keys for the radix tree.
 *
 * chunk_aligned_read() calls r5c_big_stripe_cached() to look up
 * big_stripe of each chunk in the tree. If this big_stripe is in the
 * tree, chunk_aligned_read() aborts. This look up is protected by
 * rcu_read_lock().
 *
 * It is necessary to remember whether a stripe is counted in
 * big_stripe_tree. Instead of adding new flag, we reuses existing flags:
 * STRIPE_R5C_PARTIAL_STRIPE and STRIPE_R5C_FULL_STRIPE. If either of these
 * two flags are set, the stripe is counted in big_stripe_tree. This
 * requires moving set_bit(STRIPE_R5C_PARTIAL_STRIPE) to
 * r5c_try_caching_write(); and moving clear_bit of
 * STRIPE_R5C_PARTIAL_STRIPE and STRIPE_R5C_FULL_STRIPE to
 * r5c_finish_stripe_write_out().
 */

/*
 * radix tree requests lowest 2 bits of data pointer to be 2b'00.
 * So it is necessary to left shift the counter by 2 bits before using it
 * as data pointer of the tree.
 */
#define R5C_RADIX_COUNT_SHIFT 2

/*
 * calculate key for big_stripe_tree
 *
 * sect: align_bi->bi_iter.bi_sector or sh->sector
 */
static inline sector_t r5c_tree_index(struct r5conf *conf,
				      sector_t sect)
{
	sector_t offset;

	offset = sector_div(sect, conf->chunk_sectors);
	return sect;
}

S
Shaohua Li 已提交
221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243
/*
 * 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
244
	bool need_split_bio;
S
Song Liu 已提交
245 246 247 248 249 250 251 252 253 254 255 256
	struct bio *split_bio;

	unsigned int has_flush:1;      /* include flush request */
	unsigned int has_fua:1;        /* include fua request */
	unsigned int has_null_flush:1; /* include empty flush request */
	/*
	 * io isn't sent yet, flush/fua request can only be submitted till it's
	 * the first IO in running_ios list
	 */
	unsigned int io_deferred:1;

	struct bio_list flush_barriers;   /* size == 0 flush bios */
S
Shaohua Li 已提交
257 258 259 260 261 262 263 264
};

/* 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 */
265
	IO_UNIT_STRIPE_END = 3,	/* stripes data finished writing to raid */
S
Shaohua Li 已提交
266 267
};

268 269 270 271 272 273
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
274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
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
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
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);
		}
	}
}

346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391
/* 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
 *
392 393 394 395
 * To avoid deadlock due to log space, it is necessary to reserve log
 * space to flush critical stripes (stripes that occupying log space near
 * last_checkpoint). This function helps check how much log space is
 * required to flush all cached stripes.
396
 *
397 398 399 400 401 402
 * To reduce log space requirements, two mechanisms are used to give cache
 * flush higher priorities:
 *    1. In handle_stripe_dirtying() and schedule_reconstruction(),
 *       stripes ALREADY in journal can be flushed w/o pending writes;
 *    2. In r5l_write_stripe() and r5c_cache_data(), stripes NOT in journal
 *       can be delayed (r5l_add_no_space_stripe).
403
 *
404 405 406 407 408 409 410 411 412 413 414 415
 * In cache flush, the stripe goes through 1 and then 2. For a stripe that
 * already passed 1, flushing it requires at most (conf->max_degraded + 1)
 * pages of journal space. For stripes that has not passed 1, flushing it
 * requires (conf->raid_disks + 1) pages of journal space. There are at
 * most (conf->group_cnt + 1) stripe that passed 1. So total journal space
 * required to flush all cached stripes (in pages) is:
 *
 *     (stripe_in_journal_count - group_cnt - 1) * (max_degraded + 1) +
 *     (group_cnt + 1) * (raid_disks + 1)
 * or
 *     (stripe_in_journal_count) * (max_degraded + 1) +
 *     (group_cnt + 1) * (raid_disks - max_degraded)
416 417 418 419 420 421 422 423
 */
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;

424 425 426
	return BLOCK_SECTORS *
		((conf->max_degraded + 1) * atomic_read(&log->stripe_in_journal_count) +
		 (conf->raid_disks - conf->max_degraded) * (conf->group_cnt + 1));
427 428 429 430 431 432 433 434 435 436 437 438 439 440
}

/*
 * 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;
441
	bool wake_reclaim = false;
442 443 444 445 446 447 448 449 450

	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);
451 452 453
	else {
		if (test_bit(R5C_LOG_CRITICAL, &conf->cache_state))
			wake_reclaim = true;
454
		clear_bit(R5C_LOG_CRITICAL, &conf->cache_state);
455
	}
456 457 458 459
	if (free_space < 3 * reclaim_space)
		set_bit(R5C_LOG_TIGHT, &conf->cache_state);
	else
		clear_bit(R5C_LOG_TIGHT, &conf->cache_state);
460 461 462

	if (wake_reclaim)
		r5l_wake_reclaim(log, 0);
463 464
}

465 466 467 468
/*
 * Put the stripe into writing-out phase by clearing STRIPE_R5C_CACHING.
 * This function should only be called in write-back mode.
 */
469
void r5c_make_stripe_write_out(struct stripe_head *sh)
470 471 472 473 474 475 476 477
{
	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
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505

	if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
		atomic_inc(&conf->preread_active_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);
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524
}

/*
 * 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
525 526 527 528 529 530
	} 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);
	}
531 532
}

533 534 535 536 537 538
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);
539 540 541

		r5c_finish_cache_stripe(sh);

542 543 544 545 546
		set_bit(STRIPE_HANDLE, &sh->state);
		raid5_release_stripe(sh);
	}
}

547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562
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);
	}
}

563 564 565 566 567 568 569 570 571 572 573 574 575 576
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
Song Liu 已提交
577
static void __r5l_stripe_write_finished(struct r5l_io_unit *io);
S
Shaohua Li 已提交
578 579 580
static void r5l_log_endio(struct bio *bio)
{
	struct r5l_io_unit *io = bio->bi_private;
S
Song Liu 已提交
581
	struct r5l_io_unit *io_deferred;
S
Shaohua Li 已提交
582
	struct r5l_log *log = io->log;
583
	unsigned long flags;
S
Shaohua Li 已提交
584

S
Shaohua Li 已提交
585 586 587
	if (bio->bi_error)
		md_error(log->rdev->mddev, log->rdev);

S
Shaohua Li 已提交
588
	bio_put(bio);
589
	mempool_free(io->meta_page, log->meta_pool);
S
Shaohua Li 已提交
590

591 592
	spin_lock_irqsave(&log->io_list_lock, flags);
	__r5l_set_io_unit_state(io, IO_UNIT_IO_END);
593
	if (log->need_cache_flush)
594
		r5l_move_to_end_ios(log);
595 596
	else
		r5l_log_run_stripes(log);
S
Song Liu 已提交
597 598 599 600 601 602 603 604 605 606 607
	if (!list_empty(&log->running_ios)) {
		/*
		 * FLUSH/FUA io_unit is deferred because of ordering, now we
		 * can dispatch it
		 */
		io_deferred = list_first_entry(&log->running_ios,
					       struct r5l_io_unit, log_sibling);
		if (io_deferred->io_deferred)
			schedule_work(&log->deferred_io_work);
	}

608 609
	spin_unlock_irqrestore(&log->io_list_lock, flags);

610 611
	if (log->need_cache_flush)
		md_wakeup_thread(log->rdev->mddev->thread);
S
Song Liu 已提交
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634

	if (io->has_null_flush) {
		struct bio *bi;

		WARN_ON(bio_list_empty(&io->flush_barriers));
		while ((bi = bio_list_pop(&io->flush_barriers)) != NULL) {
			bio_endio(bi);
			atomic_dec(&io->pending_stripe);
		}
		if (atomic_read(&io->pending_stripe) == 0)
			__r5l_stripe_write_finished(io);
	}
}

static void r5l_do_submit_io(struct r5l_log *log, struct r5l_io_unit *io)
{
	unsigned long flags;

	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);

	if (io->has_flush)
S
Shaohua Li 已提交
635
		io->current_bio->bi_opf |= REQ_PREFLUSH;
S
Song Liu 已提交
636
	if (io->has_fua)
S
Shaohua Li 已提交
637
		io->current_bio->bi_opf |= REQ_FUA;
S
Song Liu 已提交
638 639 640 641 642 643
	submit_bio(io->current_bio);

	if (!io->split_bio)
		return;

	if (io->has_flush)
S
Shaohua Li 已提交
644
		io->split_bio->bi_opf |= REQ_PREFLUSH;
S
Song Liu 已提交
645
	if (io->has_fua)
S
Shaohua Li 已提交
646
		io->split_bio->bi_opf |= REQ_FUA;
S
Song Liu 已提交
647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669
	submit_bio(io->split_bio);
}

/* deferred io_unit will be dispatched here */
static void r5l_submit_io_async(struct work_struct *work)
{
	struct r5l_log *log = container_of(work, struct r5l_log,
					   deferred_io_work);
	struct r5l_io_unit *io = NULL;
	unsigned long flags;

	spin_lock_irqsave(&log->io_list_lock, flags);
	if (!list_empty(&log->running_ios)) {
		io = list_first_entry(&log->running_ios, struct r5l_io_unit,
				      log_sibling);
		if (!io->io_deferred)
			io = NULL;
		else
			io->io_deferred = 0;
	}
	spin_unlock_irqrestore(&log->io_list_lock, flags);
	if (io)
		r5l_do_submit_io(log, io);
S
Shaohua Li 已提交
670 671
}

672 673 674 675 676 677 678 679 680 681 682 683 684 685 686
static void r5c_disable_writeback_async(struct work_struct *work)
{
	struct r5l_log *log = container_of(work, struct r5l_log,
					   disable_writeback_work);
	struct mddev *mddev = log->rdev->mddev;

	if (log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH)
		return;
	pr_info("md/raid:%s: Disabling writeback cache for degraded array.\n",
		mdname(mddev));
	mddev_suspend(mddev);
	log->r5c_journal_mode = R5C_JOURNAL_MODE_WRITE_THROUGH;
	mddev_resume(mddev);
}

S
Shaohua Li 已提交
687 688 689
static void r5l_submit_current_io(struct r5l_log *log)
{
	struct r5l_io_unit *io = log->current_io;
S
Song Liu 已提交
690
	struct bio *bio;
S
Shaohua Li 已提交
691
	struct r5l_meta_block *block;
692
	unsigned long flags;
S
Shaohua Li 已提交
693
	u32 crc;
S
Song Liu 已提交
694
	bool do_submit = true;
S
Shaohua Li 已提交
695 696 697 698 699 700

	if (!io)
		return;

	block = page_address(io->meta_page);
	block->meta_size = cpu_to_le32(io->meta_offset);
S
Shaohua Li 已提交
701
	crc = crc32c_le(log->uuid_checksum, block, PAGE_SIZE);
S
Shaohua Li 已提交
702
	block->checksum = cpu_to_le32(crc);
S
Song Liu 已提交
703
	bio = io->current_bio;
S
Shaohua Li 已提交
704 705

	log->current_io = NULL;
706
	spin_lock_irqsave(&log->io_list_lock, flags);
S
Song Liu 已提交
707 708 709 710 711 712 713
	if (io->has_flush || io->has_fua) {
		if (io != list_first_entry(&log->running_ios,
					   struct r5l_io_unit, log_sibling)) {
			io->io_deferred = 1;
			do_submit = false;
		}
	}
714
	spin_unlock_irqrestore(&log->io_list_lock, flags);
S
Song Liu 已提交
715 716
	if (do_submit)
		r5l_do_submit_io(log, io);
S
Shaohua Li 已提交
717 718
}

C
Christoph Hellwig 已提交
719
static struct bio *r5l_bio_alloc(struct r5l_log *log)
720
{
C
Christoph Hellwig 已提交
721
	struct bio *bio = bio_alloc_bioset(GFP_NOIO, BIO_MAX_PAGES, log->bs);
722

M
Mike Christie 已提交
723
	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
724
	bio->bi_bdev = log->rdev->bdev;
725
	bio->bi_iter.bi_sector = log->rdev->data_offset + log->log_start;
726 727 728 729

	return bio;
}

730 731 732 733
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);

734
	r5c_update_log_state(log);
735 736 737 738 739 740 741 742
	/*
	 * 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
743
		io->need_split_bio = true;
744 745 746 747

	io->log_end = log->log_start;
}

S
Shaohua Li 已提交
748 749 750 751 752
static struct r5l_io_unit *r5l_new_meta(struct r5l_log *log)
{
	struct r5l_io_unit *io;
	struct r5l_meta_block *block;

753 754 755 756 757
	io = mempool_alloc(log->io_pool, GFP_ATOMIC);
	if (!io)
		return NULL;
	memset(io, 0, sizeof(*io));

758 759 760
	io->log = log;
	INIT_LIST_HEAD(&io->log_sibling);
	INIT_LIST_HEAD(&io->stripe_list);
S
Song Liu 已提交
761
	bio_list_init(&io->flush_barriers);
762
	io->state = IO_UNIT_RUNNING;
S
Shaohua Li 已提交
763

764
	io->meta_page = mempool_alloc(log->meta_pool, GFP_NOIO);
S
Shaohua Li 已提交
765
	block = page_address(io->meta_page);
766
	clear_page(block);
S
Shaohua Li 已提交
767 768 769 770 771 772 773
	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);
774
	io->seq = log->seq++;
S
Shaohua Li 已提交
775

C
Christoph Hellwig 已提交
776 777 778
	io->current_bio = r5l_bio_alloc(log);
	io->current_bio->bi_end_io = r5l_log_endio;
	io->current_bio->bi_private = io;
779
	bio_add_page(io->current_bio, io->meta_page, PAGE_SIZE, 0);
S
Shaohua Li 已提交
780

781
	r5_reserve_log_entry(log, io);
S
Shaohua Li 已提交
782 783 784 785 786 787 788 789 790 791

	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)
{
792 793
	if (log->current_io &&
	    log->current_io->meta_offset + payload_size > PAGE_SIZE)
S
Shaohua Li 已提交
794 795
		r5l_submit_current_io(log);

796
	if (!log->current_io) {
797
		log->current_io = r5l_new_meta(log);
798 799 800 801
		if (!log->current_io)
			return -ENOMEM;
	}

S
Shaohua Li 已提交
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
	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
831
	if (io->need_split_bio) {
S
Song Liu 已提交
832 833
		BUG_ON(io->split_bio);
		io->split_bio = io->current_bio;
C
Christoph Hellwig 已提交
834
		io->current_bio = r5l_bio_alloc(log);
S
Song Liu 已提交
835 836
		bio_chain(io->current_bio, io->split_bio);
		io->need_split_bio = false;
S
Shaohua Li 已提交
837 838
	}

C
Christoph Hellwig 已提交
839 840 841
	if (!bio_add_page(io->current_bio, page, PAGE_SIZE, 0))
		BUG();

842
	r5_reserve_log_entry(log, io);
S
Shaohua Li 已提交
843 844
}

845
static int r5l_log_stripe(struct r5l_log *log, struct stripe_head *sh,
S
Shaohua Li 已提交
846 847 848 849
			   int data_pages, int parity_pages)
{
	int i;
	int meta_size;
850
	int ret;
S
Shaohua Li 已提交
851 852 853 854 855 856 857 858
	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;

859 860 861 862
	ret = r5l_get_meta(log, meta_size);
	if (ret)
		return ret;

S
Shaohua Li 已提交
863 864
	io = log->current_io;

S
Song Liu 已提交
865 866 867
	if (test_and_clear_bit(STRIPE_R5C_PREFLUSH, &sh->state))
		io->has_flush = 1;

S
Shaohua Li 已提交
868
	for (i = 0; i < sh->disks; i++) {
S
Song Liu 已提交
869 870
		if (!test_bit(R5_Wantwrite, &sh->dev[i].flags) ||
		    test_bit(R5_InJournal, &sh->dev[i].flags))
S
Shaohua Li 已提交
871 872 873
			continue;
		if (i == sh->pd_idx || i == sh->qd_idx)
			continue;
S
Song Liu 已提交
874 875 876 877 878 879 880 881 882
		if (test_bit(R5_WantFUA, &sh->dev[i].flags) &&
		    log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_BACK) {
			io->has_fua = 1;
			/*
			 * we need to flush journal to make sure recovery can
			 * reach the data with fua flag
			 */
			io->has_flush = 1;
		}
S
Shaohua Li 已提交
883 884 885 886 887 888
		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);
	}

889
	if (parity_pages == 2) {
S
Shaohua Li 已提交
890 891 892 893 894
		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);
895
	} else if (parity_pages == 1) {
S
Shaohua Li 已提交
896 897 898 899
		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);
900 901
	} else  /* Just writing data, not parity, in caching phase */
		BUG_ON(parity_pages != 0);
S
Shaohua Li 已提交
902 903 904 905

	list_add_tail(&sh->log_list, &io->stripe_list);
	atomic_inc(&io->pending_stripe);
	sh->log_io = io;
906

907 908 909 910 911 912 913 914 915 916 917 918
	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);
	}
919
	return 0;
S
Shaohua Li 已提交
920 921
}

922 923 924 925 926 927 928 929 930
/* 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
931 932 933 934 935 936
/*
 * 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)
{
937
	struct r5conf *conf = sh->raid_conf;
S
Shaohua Li 已提交
938 939 940 941
	int write_disks = 0;
	int data_pages, parity_pages;
	int reserve;
	int i;
942
	int ret = 0;
943
	bool wake_reclaim = false;
S
Shaohua Li 已提交
944 945 946 947 948 949 950 951 952 953 954

	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;
	}

955 956
	WARN_ON(test_bit(STRIPE_R5C_CACHING, &sh->state));

S
Shaohua Li 已提交
957 958 959
	for (i = 0; i < sh->disks; i++) {
		void *addr;

S
Song Liu 已提交
960 961
		if (!test_bit(R5_Wantwrite, &sh->dev[i].flags) ||
		    test_bit(R5_InJournal, &sh->dev[i].flags))
S
Shaohua Li 已提交
962
			continue;
S
Song Liu 已提交
963

S
Shaohua Li 已提交
964 965 966 967 968
		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
969 970
		sh->dev[i].log_checksum = crc32c_le(log->uuid_checksum,
						    addr, PAGE_SIZE);
S
Shaohua Li 已提交
971 972 973 974 975 976
		kunmap_atomic(addr);
	}
	parity_pages = 1 + !!(sh->qd_idx >= 0);
	data_pages = write_disks - parity_pages;

	set_bit(STRIPE_LOG_TRAPPED, &sh->state);
977 978 979 980 981
	/*
	 * 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
982 983 984 985 986 987
	atomic_inc(&sh->count);

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

988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023
	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);
			}
1024
		}
S
Shaohua Li 已提交
1025 1026
	}

1027
	mutex_unlock(&log->io_mutex);
1028 1029
	if (wake_reclaim)
		r5l_wake_reclaim(log, reserve);
S
Shaohua Li 已提交
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
	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);
}

1042 1043 1044 1045
int r5l_handle_flush_request(struct r5l_log *log, struct bio *bio)
{
	if (!log)
		return -ENODEV;
S
Song Liu 已提交
1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072

	if (log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH) {
		/*
		 * in write through (journal only)
		 * 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;
		}
		bio->bi_opf &= ~REQ_PREFLUSH;
	} else {
		/* write back (with cache) */
		if (bio->bi_iter.bi_size == 0) {
			mutex_lock(&log->io_mutex);
			r5l_get_meta(log, 0);
			bio_list_add(&log->current_io->flush_barriers, bio);
			log->current_io->has_flush = 1;
			log->current_io->has_null_flush = 1;
			atomic_inc(&log->current_io->pending_stripe);
			r5l_submit_current_io(log);
			mutex_unlock(&log->io_mutex);
			return 0;
		}
1073 1074 1075 1076
	}
	return -EAGAIN;
}

S
Shaohua Li 已提交
1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
/* 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);
}

1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110
/*
 * 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 */
1111
		spin_unlock_irqrestore(&log->stripe_in_journal_lock, flags);
1112 1113 1114 1115 1116 1117 1118 1119 1120
		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;
}

1121 1122
static sector_t r5l_reclaimable_space(struct r5l_log *log)
{
1123 1124
	struct r5conf *conf = log->rdev->mddev->private;

1125
	return r5l_ring_distance(log, log->last_checkpoint,
1126
				 r5c_calculate_new_cp(conf));
1127 1128
}

1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143
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);
	}
}

1144
static bool r5l_complete_finished_ios(struct r5l_log *log)
1145 1146 1147 1148 1149 1150
{
	struct r5l_io_unit *io, *next;
	bool found = false;

	assert_spin_locked(&log->io_list_lock);

1151
	list_for_each_entry_safe(io, next, &log->finished_ios, log_sibling) {
1152 1153 1154 1155 1156 1157 1158
		/* don't change list order */
		if (io->state < IO_UNIT_STRIPE_END)
			break;

		log->next_checkpoint = io->log_start;

		list_del(&io->log_sibling);
1159 1160
		mempool_free(io, log->io_pool);
		r5l_run_no_mem_stripe(log);
1161 1162 1163 1164 1165 1166 1167

		found = true;
	}

	return found;
}

1168 1169 1170
static void __r5l_stripe_write_finished(struct r5l_io_unit *io)
{
	struct r5l_log *log = io->log;
1171
	struct r5conf *conf = log->rdev->mddev->private;
1172 1173 1174 1175
	unsigned long flags;

	spin_lock_irqsave(&log->io_list_lock, flags);
	__r5l_set_io_unit_state(io, IO_UNIT_STRIPE_END);
1176

1177
	if (!r5l_complete_finished_ios(log)) {
1178 1179 1180
		spin_unlock_irqrestore(&log->io_list_lock, flags);
		return;
	}
1181

1182 1183
	if (r5l_reclaimable_space(log) > log->max_free_space ||
	    test_bit(R5C_LOG_TIGHT, &conf->cache_state))
1184 1185 1186 1187 1188 1189
		r5l_wake_reclaim(log, 0);

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

S
Shaohua Li 已提交
1190 1191 1192 1193 1194 1195 1196
void r5l_stripe_write_finished(struct stripe_head *sh)
{
	struct r5l_io_unit *io;

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

1197 1198
	if (io && atomic_dec_and_test(&io->pending_stripe))
		__r5l_stripe_write_finished(io);
S
Shaohua Li 已提交
1199 1200
}

1201 1202 1203 1204 1205 1206 1207
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
1208 1209 1210
	if (bio->bi_error)
		md_error(log->rdev->mddev, log->rdev);

1211
	spin_lock_irqsave(&log->io_list_lock, flags);
1212 1213
	list_for_each_entry(io, &log->flushing_ios, log_sibling)
		r5l_io_run_stripes(io);
1214
	list_splice_tail_init(&log->flushing_ios, &log->finished_ios);
1215 1216 1217
	spin_unlock_irqrestore(&log->io_list_lock, flags);
}

S
Shaohua Li 已提交
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
/*
 * 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)
{
1234
	bool do_flush;
1235 1236

	if (!log || !log->need_cache_flush)
S
Shaohua Li 已提交
1237 1238 1239
		return;

	spin_lock_irq(&log->io_list_lock);
1240 1241 1242 1243
	/* flush bio is running */
	if (!list_empty(&log->flushing_ios)) {
		spin_unlock_irq(&log->io_list_lock);
		return;
S
Shaohua Li 已提交
1244
	}
1245 1246
	list_splice_tail_init(&log->io_end_ios, &log->flushing_ios);
	do_flush = !list_empty(&log->flushing_ios);
S
Shaohua Li 已提交
1247
	spin_unlock_irq(&log->io_list_lock);
1248 1249 1250 1251 1252 1253

	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;
1254
	log->flush_bio.bi_opf = REQ_OP_WRITE | REQ_PREFLUSH;
1255
	submit_bio(&log->flush_bio);
S
Shaohua Li 已提交
1256 1257 1258
}

static void r5l_write_super(struct r5l_log *log, sector_t cp);
S
Shaohua Li 已提交
1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271
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;
	/*
1272 1273 1274 1275 1276 1277 1278 1279 1280
	 * 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
1281
	 */
1282 1283
	set_mask_bits(&mddev->sb_flags, 0,
		BIT(MD_SB_CHANGE_DEVS) | BIT(MD_SB_CHANGE_PENDING));
1284 1285 1286 1287
	if (!mddev_trylock(mddev))
		return;
	md_update_sb(mddev, 1);
	mddev_unlock(mddev);
S
Shaohua Li 已提交
1288

S
Shaohua Li 已提交
1289
	/* discard IO error really doesn't matter, ignore it */
S
Shaohua Li 已提交
1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303
	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);
	}
}

1304 1305 1306 1307 1308 1309 1310
/*
 * 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)
S
Shaohua Li 已提交
1311
{
1312 1313 1314
	BUG_ON(list_empty(&sh->lru));
	BUG_ON(!test_bit(STRIPE_R5C_CACHING, &sh->state));
	BUG_ON(test_bit(STRIPE_HANDLE, &sh->state));
S
Shaohua Li 已提交
1315 1316

	/*
1317 1318
	 * The stripe is not ON_RELEASE_LIST, so it is safe to call
	 * raid5_release_stripe() while holding conf->device_lock
S
Shaohua Li 已提交
1319
	 */
1320 1321
	BUG_ON(test_bit(STRIPE_ON_RELEASE_LIST, &sh->state));
	assert_spin_locked(&conf->device_lock);
S
Shaohua Li 已提交
1322

1323 1324
	list_del_init(&sh->lru);
	atomic_inc(&sh->count);
1325

1326 1327 1328
	set_bit(STRIPE_HANDLE, &sh->state);
	atomic_inc(&conf->active_stripes);
	r5c_make_stripe_write_out(sh);
S
Shaohua Li 已提交
1329

1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345
	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)
S
Shaohua Li 已提交
1346 1347
		return;

1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
	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);
1421 1422
				if (count++ >= R5C_RECLAIM_STRIPE_GROUP)
					break;
1423 1424 1425 1426 1427
			}
		}
		spin_unlock(&conf->device_lock);
		spin_unlock_irqrestore(&log->stripe_in_journal_lock, flags);
	}
1428 1429 1430 1431

	if (!test_bit(R5C_LOG_CRITICAL, &conf->cache_state))
		r5l_run_no_space_stripes(log);

1432 1433 1434
	md_wakeup_thread(conf->mddev->thread);
}

S
Shaohua Li 已提交
1435 1436
static void r5l_do_reclaim(struct r5l_log *log)
{
1437
	struct r5conf *conf = log->rdev->mddev->private;
S
Shaohua Li 已提交
1438
	sector_t reclaim_target = xchg(&log->reclaim_target, 0);
1439 1440
	sector_t reclaimable;
	sector_t next_checkpoint;
1441
	bool write_super;
S
Shaohua Li 已提交
1442 1443

	spin_lock_irq(&log->io_list_lock);
1444 1445
	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
1446 1447 1448 1449 1450 1451
	/*
	 * 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) {
1452 1453
		reclaimable = r5l_reclaimable_space(log);
		if (reclaimable >= reclaim_target ||
S
Shaohua Li 已提交
1454 1455
		    (list_empty(&log->running_ios) &&
		     list_empty(&log->io_end_ios) &&
1456
		     list_empty(&log->flushing_ios) &&
1457
		     list_empty(&log->finished_ios)))
S
Shaohua Li 已提交
1458 1459
			break;

1460 1461 1462 1463
		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
1464
	}
1465

1466
	next_checkpoint = r5c_calculate_new_cp(conf);
S
Shaohua Li 已提交
1467 1468
	spin_unlock_irq(&log->io_list_lock);

1469
	if (reclaimable == 0 || !write_super)
S
Shaohua Li 已提交
1470 1471 1472 1473 1474 1475 1476
		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
1477
	r5l_write_super_and_discard_space(log, next_checkpoint);
S
Shaohua Li 已提交
1478 1479

	mutex_lock(&log->io_mutex);
1480
	log->last_checkpoint = next_checkpoint;
1481
	r5c_update_log_state(log);
S
Shaohua Li 已提交
1482 1483
	mutex_unlock(&log->io_mutex);

1484
	r5l_run_no_space_stripes(log);
S
Shaohua Li 已提交
1485 1486 1487 1488 1489 1490 1491 1492 1493 1494
}

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;
1495
	r5c_do_reclaim(conf);
S
Shaohua Li 已提交
1496 1497 1498
	r5l_do_reclaim(log);
}

1499
void r5l_wake_reclaim(struct r5l_log *log, sector_t space)
S
Shaohua Li 已提交
1500
{
S
Shaohua Li 已提交
1501 1502 1503
	unsigned long target;
	unsigned long new = (unsigned long)space; /* overflow in theory */

1504 1505
	if (!log)
		return;
S
Shaohua Li 已提交
1506 1507 1508 1509 1510 1511
	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
1512 1513
}

1514 1515
void r5l_quiesce(struct r5l_log *log, int state)
{
S
Shaohua Li 已提交
1516
	struct mddev *mddev;
1517 1518
	if (!log || state == 2)
		return;
1519 1520 1521
	if (state == 0)
		kthread_unpark(log->reclaim_thread->tsk);
	else if (state == 1) {
S
Shaohua Li 已提交
1522 1523 1524
		/* make sure r5l_write_super_and_discard_space exits */
		mddev = log->rdev->mddev;
		wake_up(&mddev->sb_wait);
1525
		kthread_park(log->reclaim_thread->tsk);
1526
		r5l_wake_reclaim(log, MaxSector);
1527 1528 1529 1530
		r5l_do_reclaim(log);
	}
}

S
Shaohua Li 已提交
1531 1532
bool r5l_log_disk_error(struct r5conf *conf)
{
1533 1534
	struct r5l_log *log;
	bool ret;
1535
	/* don't allow write if journal disk is missing */
1536 1537 1538 1539 1540 1541 1542 1543 1544
	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
1545 1546
}

S
Shaohua Li 已提交
1547 1548 1549 1550 1551
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 */
S
Song Liu 已提交
1552 1553 1554
	int data_parity_stripes;	/* number of data_parity stripes */
	int data_only_stripes;		/* number of data_only stripes */
	struct list_head cached_list;
S
Shaohua Li 已提交
1555 1556
};

1557 1558
static int r5l_recovery_read_meta_block(struct r5l_log *log,
					struct r5l_recovery_ctx *ctx)
S
Shaohua Li 已提交
1559 1560 1561 1562 1563
{
	struct page *page = ctx->meta_page;
	struct r5l_meta_block *mb;
	u32 crc, stored_crc;

M
Mike Christie 已提交
1564 1565
	if (!sync_page_io(log->rdev, ctx->pos, PAGE_SIZE, page, REQ_OP_READ, 0,
			  false))
S
Shaohua Li 已提交
1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577
		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
1578
	crc = crc32c_le(log->uuid_checksum, mb, PAGE_SIZE);
S
Shaohua Li 已提交
1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589
	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;
}

1590 1591 1592 1593
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
1594 1595 1596 1597
{
	struct r5l_meta_block *mb;

	mb = page_address(page);
1598
	clear_page(mb);
S
Shaohua Li 已提交
1599 1600 1601 1602 1603
	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);
1604
}
S
Shaohua Li 已提交
1605

1606 1607 1608 1609
static int r5l_log_write_empty_meta_block(struct r5l_log *log, sector_t pos,
					  u64 seq)
{
	struct page *page;
1610
	struct r5l_meta_block *mb;
S
Shaohua Li 已提交
1611

1612 1613 1614 1615
	page = alloc_page(GFP_KERNEL);
	if (!page)
		return -ENOMEM;
	r5l_recovery_create_empty_meta_block(log, page, pos, seq);
1616 1617 1618
	mb = page_address(page);
	mb->checksum = cpu_to_le32(crc32c_le(log->uuid_checksum,
					     mb, PAGE_SIZE));
M
Mike Christie 已提交
1619
	if (!sync_page_io(log->rdev, pos, PAGE_SIZE, page, REQ_OP_WRITE,
S
Shaohua Li 已提交
1620
			  REQ_FUA, false)) {
S
Shaohua Li 已提交
1621 1622 1623 1624 1625 1626 1627
		__free_page(page);
		return -EIO;
	}
	__free_page(page);
	return 0;
}

S
Song Liu 已提交
1628 1629 1630 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
/*
 * r5l_recovery_load_data and r5l_recovery_load_parity uses flag R5_Wantwrite
 * to mark valid (potentially not flushed) data in the journal.
 *
 * We already verified checksum in r5l_recovery_verify_data_checksum_for_mb,
 * so there should not be any mismatch here.
 */
static void r5l_recovery_load_data(struct r5l_log *log,
				   struct stripe_head *sh,
				   struct r5l_recovery_ctx *ctx,
				   struct r5l_payload_data_parity *payload,
				   sector_t log_offset)
{
	struct mddev *mddev = log->rdev->mddev;
	struct r5conf *conf = mddev->private;
	int dd_idx;

	raid5_compute_sector(conf,
			     le64_to_cpu(payload->location), 0,
			     &dd_idx, sh);
	sync_page_io(log->rdev, log_offset, PAGE_SIZE,
		     sh->dev[dd_idx].page, REQ_OP_READ, 0, false);
	sh->dev[dd_idx].log_checksum =
		le32_to_cpu(payload->checksum[0]);
	ctx->meta_total_blocks += BLOCK_SECTORS;

	set_bit(R5_Wantwrite, &sh->dev[dd_idx].flags);
	set_bit(STRIPE_R5C_CACHING, &sh->state);
}

static void r5l_recovery_load_parity(struct r5l_log *log,
				     struct stripe_head *sh,
				     struct r5l_recovery_ctx *ctx,
				     struct r5l_payload_data_parity *payload,
				     sector_t log_offset)
{
	struct mddev *mddev = log->rdev->mddev;
	struct r5conf *conf = mddev->private;

	ctx->meta_total_blocks += BLOCK_SECTORS * conf->max_degraded;
	sync_page_io(log->rdev, log_offset, PAGE_SIZE,
		     sh->dev[sh->pd_idx].page, REQ_OP_READ, 0, false);
	sh->dev[sh->pd_idx].log_checksum =
		le32_to_cpu(payload->checksum[0]);
	set_bit(R5_Wantwrite, &sh->dev[sh->pd_idx].flags);

	if (sh->qd_idx >= 0) {
		sync_page_io(log->rdev,
			     r5l_ring_add(log, log_offset, BLOCK_SECTORS),
			     PAGE_SIZE, sh->dev[sh->qd_idx].page,
			     REQ_OP_READ, 0, false);
		sh->dev[sh->qd_idx].log_checksum =
			le32_to_cpu(payload->checksum[1]);
		set_bit(R5_Wantwrite, &sh->dev[sh->qd_idx].flags);
S
Shaohua Li 已提交
1682
	}
S
Song Liu 已提交
1683 1684
	clear_bit(STRIPE_R5C_CACHING, &sh->state);
}
S
Shaohua Li 已提交
1685

S
Song Liu 已提交
1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703
static void r5l_recovery_reset_stripe(struct stripe_head *sh)
{
	int i;

	sh->state = 0;
	sh->log_start = MaxSector;
	for (i = sh->disks; i--; )
		sh->dev[i].flags = 0;
}

static void
r5l_recovery_replay_one_stripe(struct r5conf *conf,
			       struct stripe_head *sh,
			       struct r5l_recovery_ctx *ctx)
{
	struct md_rdev *rdev, *rrdev;
	int disk_index;
	int data_count = 0;
S
Shaohua Li 已提交
1704

S
Song Liu 已提交
1705
	for (disk_index = 0; disk_index < sh->disks; disk_index++) {
S
Shaohua Li 已提交
1706 1707
		if (!test_bit(R5_Wantwrite, &sh->dev[disk_index].flags))
			continue;
S
Song Liu 已提交
1708 1709 1710
		if (disk_index == sh->qd_idx || disk_index == sh->pd_idx)
			continue;
		data_count++;
S
Shaohua Li 已提交
1711 1712
	}

S
Song Liu 已提交
1713 1714 1715 1716 1717 1718 1719
	/*
	 * stripes that only have parity must have been flushed
	 * before the crash that we are now recovering from, so
	 * there is nothing more to recovery.
	 */
	if (data_count == 0)
		goto out;
S
Shaohua Li 已提交
1720

S
Song Liu 已提交
1721 1722
	for (disk_index = 0; disk_index < sh->disks; disk_index++) {
		if (!test_bit(R5_Wantwrite, &sh->dev[disk_index].flags))
S
Shaohua Li 已提交
1723 1724 1725
			continue;

		/* in case device is broken */
S
Song Liu 已提交
1726
		rcu_read_lock();
S
Shaohua Li 已提交
1727
		rdev = rcu_dereference(conf->disks[disk_index].rdev);
S
Song Liu 已提交
1728 1729 1730 1731
		if (rdev) {
			atomic_inc(&rdev->nr_pending);
			rcu_read_unlock();
			sync_page_io(rdev, sh->sector, PAGE_SIZE,
M
Mike Christie 已提交
1732 1733
				     sh->dev[disk_index].page, REQ_OP_WRITE, 0,
				     false);
S
Song Liu 已提交
1734 1735 1736
			rdev_dec_pending(rdev, rdev->mddev);
			rcu_read_lock();
		}
S
Shaohua Li 已提交
1737
		rrdev = rcu_dereference(conf->disks[disk_index].replacement);
S
Song Liu 已提交
1738 1739 1740 1741
		if (rrdev) {
			atomic_inc(&rrdev->nr_pending);
			rcu_read_unlock();
			sync_page_io(rrdev, sh->sector, PAGE_SIZE,
M
Mike Christie 已提交
1742 1743
				     sh->dev[disk_index].page, REQ_OP_WRITE, 0,
				     false);
S
Song Liu 已提交
1744 1745 1746 1747
			rdev_dec_pending(rrdev, rrdev->mddev);
			rcu_read_lock();
		}
		rcu_read_unlock();
S
Shaohua Li 已提交
1748
	}
S
Song Liu 已提交
1749 1750 1751 1752 1753 1754 1755
	ctx->data_parity_stripes++;
out:
	r5l_recovery_reset_stripe(sh);
}

static struct stripe_head *
r5c_recovery_alloc_stripe(struct r5conf *conf,
1756
			  sector_t stripe_sect)
S
Song Liu 已提交
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 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 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 1870 1871 1872 1873 1874
{
	struct stripe_head *sh;

	sh = raid5_get_active_stripe(conf, stripe_sect, 0, 1, 0);
	if (!sh)
		return NULL;  /* no more stripe available */

	r5l_recovery_reset_stripe(sh);

	return sh;
}

static struct stripe_head *
r5c_recovery_lookup_stripe(struct list_head *list, sector_t sect)
{
	struct stripe_head *sh;

	list_for_each_entry(sh, list, lru)
		if (sh->sector == sect)
			return sh;
	return NULL;
}

static void
r5c_recovery_drop_stripes(struct list_head *cached_stripe_list,
			  struct r5l_recovery_ctx *ctx)
{
	struct stripe_head *sh, *next;

	list_for_each_entry_safe(sh, next, cached_stripe_list, lru) {
		r5l_recovery_reset_stripe(sh);
		list_del_init(&sh->lru);
		raid5_release_stripe(sh);
	}
}

static void
r5c_recovery_replay_stripes(struct list_head *cached_stripe_list,
			    struct r5l_recovery_ctx *ctx)
{
	struct stripe_head *sh, *next;

	list_for_each_entry_safe(sh, next, cached_stripe_list, lru)
		if (!test_bit(STRIPE_R5C_CACHING, &sh->state)) {
			r5l_recovery_replay_one_stripe(sh->raid_conf, sh, ctx);
			list_del_init(&sh->lru);
			raid5_release_stripe(sh);
		}
}

/* if matches return 0; otherwise return -EINVAL */
static int
r5l_recovery_verify_data_checksum(struct r5l_log *log, struct page *page,
				  sector_t log_offset, __le32 log_checksum)
{
	void *addr;
	u32 checksum;

	sync_page_io(log->rdev, log_offset, PAGE_SIZE,
		     page, REQ_OP_READ, 0, false);
	addr = kmap_atomic(page);
	checksum = crc32c_le(log->uuid_checksum, addr, PAGE_SIZE);
	kunmap_atomic(addr);
	return (le32_to_cpu(log_checksum) == checksum) ? 0 : -EINVAL;
}

/*
 * before loading data to stripe cache, we need verify checksum for all data,
 * if there is mismatch for any data page, we drop all data in the mata block
 */
static int
r5l_recovery_verify_data_checksum_for_mb(struct r5l_log *log,
					 struct r5l_recovery_ctx *ctx)
{
	struct mddev *mddev = log->rdev->mddev;
	struct r5conf *conf = mddev->private;
	struct r5l_meta_block *mb = page_address(ctx->meta_page);
	sector_t mb_offset = sizeof(struct r5l_meta_block);
	sector_t log_offset = r5l_ring_add(log, ctx->pos, BLOCK_SECTORS);
	struct page *page;
	struct r5l_payload_data_parity *payload;

	page = alloc_page(GFP_KERNEL);
	if (!page)
		return -ENOMEM;

	while (mb_offset < le32_to_cpu(mb->meta_size)) {
		payload = (void *)mb + mb_offset;

		if (payload->header.type == R5LOG_PAYLOAD_DATA) {
			if (r5l_recovery_verify_data_checksum(
				    log, page, log_offset,
				    payload->checksum[0]) < 0)
				goto mismatch;
		} else if (payload->header.type == R5LOG_PAYLOAD_PARITY) {
			if (r5l_recovery_verify_data_checksum(
				    log, page, log_offset,
				    payload->checksum[0]) < 0)
				goto mismatch;
			if (conf->max_degraded == 2 && /* q for RAID 6 */
			    r5l_recovery_verify_data_checksum(
				    log, page,
				    r5l_ring_add(log, log_offset,
						 BLOCK_SECTORS),
				    payload->checksum[1]) < 0)
				goto mismatch;
		} else /* not R5LOG_PAYLOAD_DATA or R5LOG_PAYLOAD_PARITY */
			goto mismatch;

		log_offset = r5l_ring_add(log, log_offset,
					  le32_to_cpu(payload->size));

		mb_offset += sizeof(struct r5l_payload_data_parity) +
			sizeof(__le32) *
			(le32_to_cpu(payload->size) >> (PAGE_SHIFT - 9));
	}

	put_page(page);
S
Shaohua Li 已提交
1875 1876
	return 0;

S
Song Liu 已提交
1877 1878
mismatch:
	put_page(page);
S
Shaohua Li 已提交
1879 1880 1881
	return -EINVAL;
}

S
Song Liu 已提交
1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893
/*
 * Analyze all data/parity pages in one meta block
 * Returns:
 * 0 for success
 * -EINVAL for unknown playload type
 * -EAGAIN for checksum mismatch of data page
 * -ENOMEM for run out of memory (alloc_page failed or run out of stripes)
 */
static int
r5c_recovery_analyze_meta_block(struct r5l_log *log,
				struct r5l_recovery_ctx *ctx,
				struct list_head *cached_stripe_list)
S
Shaohua Li 已提交
1894
{
S
Song Liu 已提交
1895 1896
	struct mddev *mddev = log->rdev->mddev;
	struct r5conf *conf = mddev->private;
S
Shaohua Li 已提交
1897
	struct r5l_meta_block *mb;
S
Song Liu 已提交
1898 1899
	struct r5l_payload_data_parity *payload;
	int mb_offset;
S
Shaohua Li 已提交
1900
	sector_t log_offset;
S
Song Liu 已提交
1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914
	sector_t stripe_sect;
	struct stripe_head *sh;
	int ret;

	/*
	 * for mismatch in data blocks, we will drop all data in this mb, but
	 * we will still read next mb for other data with FLUSH flag, as
	 * io_unit could finish out of order.
	 */
	ret = r5l_recovery_verify_data_checksum_for_mb(log, ctx);
	if (ret == -EINVAL)
		return -EAGAIN;
	else if (ret)
		return ret;   /* -ENOMEM duo to alloc_page() failed */
S
Shaohua Li 已提交
1915 1916

	mb = page_address(ctx->meta_page);
S
Song Liu 已提交
1917
	mb_offset = sizeof(struct r5l_meta_block);
S
Shaohua Li 已提交
1918 1919
	log_offset = r5l_ring_add(log, ctx->pos, BLOCK_SECTORS);

S
Song Liu 已提交
1920
	while (mb_offset < le32_to_cpu(mb->meta_size)) {
S
Shaohua Li 已提交
1921 1922
		int dd;

S
Song Liu 已提交
1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933
		payload = (void *)mb + mb_offset;
		stripe_sect = (payload->header.type == R5LOG_PAYLOAD_DATA) ?
			raid5_compute_sector(
				conf, le64_to_cpu(payload->location), 0, &dd,
				NULL)
			: le64_to_cpu(payload->location);

		sh = r5c_recovery_lookup_stripe(cached_stripe_list,
						stripe_sect);

		if (!sh) {
1934
			sh = r5c_recovery_alloc_stripe(conf, stripe_sect);
S
Song Liu 已提交
1935 1936 1937 1938 1939 1940 1941 1942
			/*
			 * cannot get stripe from raid5_get_active_stripe
			 * try replay some stripes
			 */
			if (!sh) {
				r5c_recovery_replay_stripes(
					cached_stripe_list, ctx);
				sh = r5c_recovery_alloc_stripe(
1943
					conf, stripe_sect);
S
Song Liu 已提交
1944 1945 1946 1947 1948 1949 1950
			}
			if (!sh) {
				pr_debug("md/raid:%s: Increasing stripe cache size to %d to recovery data on journal.\n",
					mdname(mddev),
					conf->min_nr_stripes * 2);
				raid5_set_cache_size(mddev,
						     conf->min_nr_stripes * 2);
1951 1952
				sh = r5c_recovery_alloc_stripe(conf,
							       stripe_sect);
S
Song Liu 已提交
1953 1954 1955 1956 1957 1958 1959 1960 1961 1962
			}
			if (!sh) {
				pr_err("md/raid:%s: Cannot get enough stripes due to memory pressure. Recovery failed.\n",
				       mdname(mddev));
				return -ENOMEM;
			}
			list_add_tail(&sh->lru, cached_stripe_list);
		}

		if (payload->header.type == R5LOG_PAYLOAD_DATA) {
1963 1964
			if (!test_bit(STRIPE_R5C_CACHING, &sh->state) &&
			    test_bit(R5_Wantwrite, &sh->dev[sh->pd_idx].flags)) {
S
Song Liu 已提交
1965 1966 1967 1968 1969 1970 1971 1972 1973
				r5l_recovery_replay_one_stripe(conf, sh, ctx);
				list_move_tail(&sh->lru, cached_stripe_list);
			}
			r5l_recovery_load_data(log, sh, ctx, payload,
					       log_offset);
		} else if (payload->header.type == R5LOG_PAYLOAD_PARITY)
			r5l_recovery_load_parity(log, sh, ctx, payload,
						 log_offset);
		else
S
Shaohua Li 已提交
1974
			return -EINVAL;
S
Song Liu 已提交
1975 1976 1977 1978 1979 1980 1981

		log_offset = r5l_ring_add(log, log_offset,
					  le32_to_cpu(payload->size));

		mb_offset += sizeof(struct r5l_payload_data_parity) +
			sizeof(__le32) *
			(le32_to_cpu(payload->size) >> (PAGE_SHIFT - 9));
S
Shaohua Li 已提交
1982
	}
S
Song Liu 已提交
1983

S
Shaohua Li 已提交
1984 1985 1986
	return 0;
}

S
Song Liu 已提交
1987 1988 1989 1990 1991 1992
/*
 * Load the stripe into cache. The stripe will be written out later by
 * the stripe cache state machine.
 */
static void r5c_recovery_load_one_stripe(struct r5l_log *log,
					 struct stripe_head *sh)
S
Shaohua Li 已提交
1993
{
S
Song Liu 已提交
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
	struct r5dev *dev;
	int i;

	for (i = sh->disks; i--; ) {
		dev = sh->dev + i;
		if (test_and_clear_bit(R5_Wantwrite, &dev->flags)) {
			set_bit(R5_InJournal, &dev->flags);
			set_bit(R5_UPTODATE, &dev->flags);
		}
	}
}

/*
 * Scan through the log for all to-be-flushed data
 *
 * For stripes with data and parity, namely Data-Parity stripe
 * (STRIPE_R5C_CACHING == 0), we simply replay all the writes.
 *
 * For stripes with only data, namely Data-Only stripe
 * (STRIPE_R5C_CACHING == 1), we load them to stripe cache state machine.
 *
 * For a stripe, if we see data after parity, we should discard all previous
 * data and parity for this stripe, as these data are already flushed to
 * the array.
 *
 * At the end of the scan, we return the new journal_tail, which points to
 * first data-only stripe on the journal device, or next invalid meta block.
 */
static int r5c_recovery_flush_log(struct r5l_log *log,
				  struct r5l_recovery_ctx *ctx)
{
2025
	struct stripe_head *sh;
S
Song Liu 已提交
2026 2027 2028
	int ret = 0;

	/* scan through the log */
S
Shaohua Li 已提交
2029
	while (1) {
S
Song Liu 已提交
2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040
		if (r5l_recovery_read_meta_block(log, ctx))
			break;

		ret = r5c_recovery_analyze_meta_block(log, ctx,
						      &ctx->cached_list);
		/*
		 * -EAGAIN means mismatch in data block, in this case, we still
		 * try scan the next metablock
		 */
		if (ret && ret != -EAGAIN)
			break;   /* ret == -EINVAL or -ENOMEM */
S
Shaohua Li 已提交
2041 2042 2043
		ctx->seq++;
		ctx->pos = r5l_ring_add(log, ctx->pos, ctx->meta_total_blocks);
	}
S
Song Liu 已提交
2044 2045 2046 2047 2048 2049 2050 2051 2052 2053

	if (ret == -ENOMEM) {
		r5c_recovery_drop_stripes(&ctx->cached_list, ctx);
		return ret;
	}

	/* replay data-parity stripes */
	r5c_recovery_replay_stripes(&ctx->cached_list, ctx);

	/* load data-only stripes to stripe cache */
2054
	list_for_each_entry(sh, &ctx->cached_list, lru) {
S
Song Liu 已提交
2055 2056 2057 2058 2059 2060
		WARN_ON(!test_bit(STRIPE_R5C_CACHING, &sh->state));
		r5c_recovery_load_one_stripe(log, sh);
		ctx->data_only_stripes++;
	}

	return 0;
S
Shaohua Li 已提交
2061 2062
}

S
Song Liu 已提交
2063 2064 2065 2066 2067 2068 2069 2070 2071 2072
/*
 * 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
2073 2074
 * 1's seq + 10000 and let superblock points to meta2. The same recovery
 * will not think meta 3 is a valid meta, because its seq doesn't match
S
Song Liu 已提交
2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103
 */

/*
 * Before recovery, the log looks like the following
 *
 *   ---------------------------------------------
 *   |           valid log        | invalid log  |
 *   ---------------------------------------------
 *   ^
 *   |- log->last_checkpoint
 *   |- log->last_cp_seq
 *
 * Now we scan through the log until we see invalid entry
 *
 *   ---------------------------------------------
 *   |           valid log        | invalid log  |
 *   ---------------------------------------------
 *   ^                            ^
 *   |- log->last_checkpoint      |- ctx->pos
 *   |- log->last_cp_seq          |- ctx->seq
 *
 * From this point, we need to increase seq number by 10 to avoid
 * confusing next recovery.
 *
 *   ---------------------------------------------
 *   |           valid log        | invalid log  |
 *   ---------------------------------------------
 *   ^                              ^
 *   |- log->last_checkpoint        |- ctx->pos+1
2104
 *   |- log->last_cp_seq            |- ctx->seq+10001
S
Song Liu 已提交
2105 2106 2107 2108 2109 2110 2111 2112 2113 2114
 *
 * However, it is not safe to start the state machine yet, because data only
 * parities are not yet secured in RAID. To save these data only parities, we
 * rewrite them from seq+11.
 *
 *   -----------------------------------------------------------------
 *   |           valid log        | data only stripes | invalid log  |
 *   -----------------------------------------------------------------
 *   ^                                                ^
 *   |- log->last_checkpoint                          |- ctx->pos+n
2115
 *   |- log->last_cp_seq                              |- ctx->seq+10000+n
S
Song Liu 已提交
2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126
 *
 * If failure happens again during this process, the recovery can safe start
 * again from log->last_checkpoint.
 *
 * Once data only stripes are rewritten to journal, we move log_tail
 *
 *   -----------------------------------------------------------------
 *   |     old log        |    data only stripes    | invalid log  |
 *   -----------------------------------------------------------------
 *                        ^                         ^
 *                        |- log->last_checkpoint   |- ctx->pos+n
2127
 *                        |- log->last_cp_seq       |- ctx->seq+10000+n
S
Song Liu 已提交
2128 2129 2130 2131 2132 2133 2134
 *
 * Then we can safely start the state machine. If failure happens from this
 * point on, the recovery will start from new log->last_checkpoint.
 */
static int
r5c_recovery_rewrite_data_only_stripes(struct r5l_log *log,
				       struct r5l_recovery_ctx *ctx)
S
Shaohua Li 已提交
2135
{
2136
	struct stripe_head *sh;
S
Song Liu 已提交
2137
	struct mddev *mddev = log->rdev->mddev;
S
Shaohua Li 已提交
2138
	struct page *page;
2139
	sector_t next_checkpoint = MaxSector;
S
Shaohua Li 已提交
2140

S
Song Liu 已提交
2141 2142 2143 2144
	page = alloc_page(GFP_KERNEL);
	if (!page) {
		pr_err("md/raid:%s: cannot allocate memory to rewrite data only stripes\n",
		       mdname(mddev));
S
Shaohua Li 已提交
2145
		return -ENOMEM;
S
Song Liu 已提交
2146
	}
S
Shaohua Li 已提交
2147

2148 2149
	WARN_ON(list_empty(&ctx->cached_list));

2150
	list_for_each_entry(sh, &ctx->cached_list, lru) {
S
Song Liu 已提交
2151 2152 2153 2154 2155 2156 2157 2158 2159 2160
		struct r5l_meta_block *mb;
		int i;
		int offset;
		sector_t write_pos;

		WARN_ON(!test_bit(STRIPE_R5C_CACHING, &sh->state));
		r5l_recovery_create_empty_meta_block(log, page,
						     ctx->pos, ctx->seq);
		mb = page_address(page);
		offset = le32_to_cpu(mb->meta_size);
2161
		write_pos = r5l_ring_add(log, ctx->pos, BLOCK_SECTORS);
S
Song Liu 已提交
2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189

		for (i = sh->disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			struct r5l_payload_data_parity *payload;
			void *addr;

			if (test_bit(R5_InJournal, &dev->flags)) {
				payload = (void *)mb + offset;
				payload->header.type = cpu_to_le16(
					R5LOG_PAYLOAD_DATA);
				payload->size = BLOCK_SECTORS;
				payload->location = cpu_to_le64(
					raid5_compute_blocknr(sh, i, 0));
				addr = kmap_atomic(dev->page);
				payload->checksum[0] = cpu_to_le32(
					crc32c_le(log->uuid_checksum, addr,
						  PAGE_SIZE));
				kunmap_atomic(addr);
				sync_page_io(log->rdev, write_pos, PAGE_SIZE,
					     dev->page, REQ_OP_WRITE, 0, false);
				write_pos = r5l_ring_add(log, write_pos,
							 BLOCK_SECTORS);
				offset += sizeof(__le32) +
					sizeof(struct r5l_payload_data_parity);

			}
		}
		mb->meta_size = cpu_to_le32(offset);
2190 2191
		mb->checksum = cpu_to_le32(crc32c_le(log->uuid_checksum,
						     mb, PAGE_SIZE));
S
Song Liu 已提交
2192
		sync_page_io(log->rdev, ctx->pos, PAGE_SIZE, page,
S
Shaohua Li 已提交
2193
			     REQ_OP_WRITE, REQ_FUA, false);
S
Song Liu 已提交
2194
		sh->log_start = ctx->pos;
2195 2196
		list_add_tail(&sh->r5c, &log->stripe_in_journal_list);
		atomic_inc(&log->stripe_in_journal_count);
S
Song Liu 已提交
2197 2198
		ctx->pos = write_pos;
		ctx->seq += 1;
2199
		next_checkpoint = sh->log_start;
S
Shaohua Li 已提交
2200
	}
2201
	log->next_checkpoint = next_checkpoint;
S
Shaohua Li 已提交
2202 2203 2204 2205
	__free_page(page);
	return 0;
}

2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232
static void r5c_recovery_flush_data_only_stripes(struct r5l_log *log,
						 struct r5l_recovery_ctx *ctx)
{
	struct mddev *mddev = log->rdev->mddev;
	struct r5conf *conf = mddev->private;
	struct stripe_head *sh, *next;

	if (ctx->data_only_stripes == 0)
		return;

	log->r5c_journal_mode = R5C_JOURNAL_MODE_WRITE_BACK;

	list_for_each_entry_safe(sh, next, &ctx->cached_list, lru) {
		r5c_make_stripe_write_out(sh);
		set_bit(STRIPE_HANDLE, &sh->state);
		list_del_init(&sh->lru);
		raid5_release_stripe(sh);
	}

	md_wakeup_thread(conf->mddev->thread);
	/* reuse conf->wait_for_quiescent in recovery */
	wait_event(conf->wait_for_quiescent,
		   atomic_read(&conf->active_stripes) == 0);

	log->r5c_journal_mode = R5C_JOURNAL_MODE_WRITE_THROUGH;
}

S
Shaohua Li 已提交
2233 2234
static int r5l_recovery_log(struct r5l_log *log)
{
S
Song Liu 已提交
2235
	struct mddev *mddev = log->rdev->mddev;
S
Shaohua Li 已提交
2236
	struct r5l_recovery_ctx ctx;
S
Song Liu 已提交
2237
	int ret;
2238
	sector_t pos;
S
Shaohua Li 已提交
2239 2240 2241 2242

	ctx.pos = log->last_checkpoint;
	ctx.seq = log->last_cp_seq;
	ctx.meta_page = alloc_page(GFP_KERNEL);
S
Song Liu 已提交
2243 2244 2245 2246
	ctx.data_only_stripes = 0;
	ctx.data_parity_stripes = 0;
	INIT_LIST_HEAD(&ctx.cached_list);

S
Shaohua Li 已提交
2247 2248 2249
	if (!ctx.meta_page)
		return -ENOMEM;

S
Song Liu 已提交
2250
	ret = r5c_recovery_flush_log(log, &ctx);
S
Shaohua Li 已提交
2251 2252
	__free_page(ctx.meta_page);

S
Song Liu 已提交
2253 2254
	if (ret)
		return ret;
S
Song Liu 已提交
2255

2256 2257
	pos = ctx.pos;
	ctx.seq += 10000;
2258 2259


S
Song Liu 已提交
2260 2261 2262
	if ((ctx.data_only_stripes == 0) && (ctx.data_parity_stripes == 0))
		pr_debug("md/raid:%s: starting from clean shutdown\n",
			 mdname(mddev));
2263
	else
2264
		pr_debug("md/raid:%s: recovering %d data-only stripes and %d data-parity stripes\n",
S
Song Liu 已提交
2265 2266 2267
			 mdname(mddev), ctx.data_only_stripes,
			 ctx.data_parity_stripes);

2268 2269 2270 2271 2272 2273 2274 2275
	if (ctx.data_only_stripes == 0) {
		log->next_checkpoint = ctx.pos;
		r5l_log_write_empty_meta_block(log, ctx.pos, ctx.seq++);
		ctx.pos = r5l_ring_add(log, ctx.pos, BLOCK_SECTORS);
	} else if (r5c_recovery_rewrite_data_only_stripes(log, &ctx)) {
		pr_err("md/raid:%s: failed to rewrite stripes to journal\n",
		       mdname(mddev));
		return -EIO;
S
Song Liu 已提交
2276 2277
	}

S
Song Liu 已提交
2278 2279
	log->log_start = ctx.pos;
	log->seq = ctx.seq;
2280 2281
	log->last_checkpoint = pos;
	r5l_write_super(log, pos);
2282 2283

	r5c_recovery_flush_data_only_stripes(log, &ctx);
S
Shaohua Li 已提交
2284 2285 2286 2287 2288 2289 2290 2291
	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;
2292
	set_bit(MD_SB_CHANGE_DEVS, &mddev->sb_flags);
S
Shaohua Li 已提交
2293 2294
}

S
Song Liu 已提交
2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344
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;

2345 2346 2347 2348
	if (raid5_calc_degraded(conf) > 0 &&
	    val == R5C_JOURNAL_MODE_WRITE_BACK)
		return -EINVAL;

S
Song Liu 已提交
2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361
	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);

2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375
/*
 * 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
2376 2377 2378
	int i;
	struct r5dev *dev;
	int to_cache = 0;
2379 2380 2381 2382
	void **pslot;
	sector_t tree_index;
	int ret;
	uintptr_t refcount;
2383 2384 2385

	BUG_ON(!r5c_is_writeback(log));

S
Song Liu 已提交
2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406
	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);
	}

2407 2408 2409 2410 2411 2412 2413 2414 2415 2416
	/*
	 * When run in degraded mode, array is set to write-through mode.
	 * This check helps drain pending write safely in the transition to
	 * write-through mode.
	 */
	if (s->failed) {
		r5c_make_stripe_write_out(sh);
		return -EAGAIN;
	}

S
Song Liu 已提交
2417 2418 2419 2420 2421 2422 2423 2424 2425 2426
	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;
		}
	}

2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464
	/* if the stripe is not counted in big_stripe_tree, add it now */
	if (!test_bit(STRIPE_R5C_PARTIAL_STRIPE, &sh->state) &&
	    !test_bit(STRIPE_R5C_FULL_STRIPE, &sh->state)) {
		tree_index = r5c_tree_index(conf, sh->sector);
		spin_lock(&log->tree_lock);
		pslot = radix_tree_lookup_slot(&log->big_stripe_tree,
					       tree_index);
		if (pslot) {
			refcount = (uintptr_t)radix_tree_deref_slot_protected(
				pslot, &log->tree_lock) >>
				R5C_RADIX_COUNT_SHIFT;
			radix_tree_replace_slot(
				&log->big_stripe_tree, pslot,
				(void *)((refcount + 1) << R5C_RADIX_COUNT_SHIFT));
		} else {
			/*
			 * this radix_tree_insert can fail safely, so no
			 * need to call radix_tree_preload()
			 */
			ret = radix_tree_insert(
				&log->big_stripe_tree, tree_index,
				(void *)(1 << R5C_RADIX_COUNT_SHIFT));
			if (ret) {
				spin_unlock(&log->tree_lock);
				r5c_make_stripe_write_out(sh);
				return -EAGAIN;
			}
		}
		spin_unlock(&log->tree_lock);

		/*
		 * set STRIPE_R5C_PARTIAL_STRIPE, this shows the stripe is
		 * counted in the radix tree
		 */
		set_bit(STRIPE_R5C_PARTIAL_STRIPE, &sh->state);
		atomic_inc(&conf->r5c_cached_partial_stripes);
	}

S
Song Liu 已提交
2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492
	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)
{
S
Song Liu 已提交
2493
	struct r5conf *conf = sh->raid_conf;
S
Song Liu 已提交
2494
	int i;
S
Song Liu 已提交
2495 2496 2497 2498
	bool using_disk_info_extra_page;

	using_disk_info_extra_page =
		sh->dev[0].orig_page == conf->disks[0].extra_page;
S
Song Liu 已提交
2499 2500 2501 2502 2503 2504

	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;
2505 2506
			clear_bit(R5_OrigPageUPTDODATE, &sh->dev[i].flags);

S
Song Liu 已提交
2507 2508
			if (!using_disk_info_extra_page)
				put_page(p);
S
Song Liu 已提交
2509
		}
S
Song Liu 已提交
2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528

	if (using_disk_info_extra_page) {
		clear_bit(R5C_EXTRA_PAGE_IN_USE, &conf->cache_state);
		md_wakeup_thread(conf->mddev->thread);
	}
}

void r5c_use_extra_page(struct stripe_head *sh)
{
	struct r5conf *conf = sh->raid_conf;
	int i;
	struct r5dev *dev;

	for (i = sh->disks; i--; ) {
		dev = &sh->dev[i];
		if (dev->orig_page != dev->page)
			put_page(dev->orig_page);
		dev->orig_page = conf->disks[i].extra_page;
	}
2529 2530 2531 2532 2533 2534 2535 2536 2537 2538
}

/*
 * 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)
{
2539
	struct r5l_log *log = conf->log;
S
Song Liu 已提交
2540 2541
	int i;
	int do_wakeup = 0;
2542 2543 2544
	sector_t tree_index;
	void **pslot;
	uintptr_t refcount;
S
Song Liu 已提交
2545

2546
	if (!log || !test_bit(R5_InJournal, &sh->dev[sh->pd_idx].flags))
2547 2548 2549 2550 2551
		return;

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

2552
	if (log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_THROUGH)
2553
		return;
S
Song Liu 已提交
2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572

	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);
2573

2574
	spin_lock_irq(&log->stripe_in_journal_lock);
2575
	list_del_init(&sh->r5c);
2576
	spin_unlock_irq(&log->stripe_in_journal_lock);
2577
	sh->log_start = MaxSector;
2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610

	atomic_dec(&log->stripe_in_journal_count);
	r5c_update_log_state(log);

	/* stop counting this stripe in big_stripe_tree */
	if (test_bit(STRIPE_R5C_PARTIAL_STRIPE, &sh->state) ||
	    test_bit(STRIPE_R5C_FULL_STRIPE, &sh->state)) {
		tree_index = r5c_tree_index(conf, sh->sector);
		spin_lock(&log->tree_lock);
		pslot = radix_tree_lookup_slot(&log->big_stripe_tree,
					       tree_index);
		BUG_ON(pslot == NULL);
		refcount = (uintptr_t)radix_tree_deref_slot_protected(
			pslot, &log->tree_lock) >>
			R5C_RADIX_COUNT_SHIFT;
		if (refcount == 1)
			radix_tree_delete(&log->big_stripe_tree, tree_index);
		else
			radix_tree_replace_slot(
				&log->big_stripe_tree, pslot,
				(void *)((refcount - 1) << R5C_RADIX_COUNT_SHIFT));
		spin_unlock(&log->tree_lock);
	}

	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);
	}
S
Song Liu 已提交
2611 2612 2613 2614 2615 2616
}

int
r5c_cache_data(struct r5l_log *log, struct stripe_head *sh,
	       struct stripe_head_state *s)
{
2617
	struct r5conf *conf = sh->raid_conf;
S
Song Liu 已提交
2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648
	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);

2649 2650 2651 2652 2653 2654 2655 2656
	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
2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667
	} 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;
S
Shaohua Li 已提交
2668 2669
}

2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685
/* check whether this big stripe is in write back cache. */
bool r5c_big_stripe_cached(struct r5conf *conf, sector_t sect)
{
	struct r5l_log *log = conf->log;
	sector_t tree_index;
	void *slot;

	if (!log)
		return false;

	WARN_ON_ONCE(!rcu_read_lock_held());
	tree_index = r5c_tree_index(conf, sect);
	slot = radix_tree_lookup(&log->big_stripe_tree, tree_index);
	return slot != NULL;
}

S
Shaohua Li 已提交
2686 2687 2688 2689 2690 2691 2692 2693
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;
2694
	int ret = 0;
S
Shaohua Li 已提交
2695 2696 2697 2698 2699 2700 2701 2702

	/* 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
2703
	if (!sync_page_io(rdev, cp, PAGE_SIZE, page, REQ_OP_READ, 0, false)) {
S
Shaohua Li 已提交
2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715
		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
2716
	expected_crc = crc32c_le(log->uuid_checksum, mb, PAGE_SIZE);
S
Shaohua Li 已提交
2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728
	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;
2729
		r5l_log_write_empty_meta_block(log, cp, log->last_cp_seq);
S
Shaohua Li 已提交
2730 2731 2732 2733 2734 2735 2736 2737 2738 2739
		/*
		 * 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
2740 2741 2742
	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
2743 2744 2745 2746
	log->last_checkpoint = cp;

	__free_page(page);

2747 2748 2749 2750 2751 2752 2753
	if (create_super) {
		log->log_start = r5l_ring_add(log, cp, BLOCK_SECTORS);
		log->seq = log->last_cp_seq + 1;
		log->next_checkpoint = cp;
	} else
		ret = r5l_recovery_log(log);

2754 2755
	r5c_update_log_state(log);
	return ret;
S
Shaohua Li 已提交
2756 2757 2758 2759 2760
ioerr:
	__free_page(page);
	return ret;
}

2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773
void r5c_update_on_rdev_error(struct mddev *mddev)
{
	struct r5conf *conf = mddev->private;
	struct r5l_log *log = conf->log;

	if (!log)
		return;

	if (raid5_calc_degraded(conf) > 0 &&
	    conf->log->r5c_journal_mode == R5C_JOURNAL_MODE_WRITE_BACK)
		schedule_work(&log->disable_writeback_work);
}

S
Shaohua Li 已提交
2774 2775
int r5l_init_log(struct r5conf *conf, struct md_rdev *rdev)
{
J
Jens Axboe 已提交
2776
	struct request_queue *q = bdev_get_queue(rdev->bdev);
S
Shaohua Li 已提交
2777 2778 2779 2780
	struct r5l_log *log;

	if (PAGE_SIZE != 4096)
		return -EINVAL;
2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796

	/*
	 * 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
2797 2798 2799 2800 2801
	log = kzalloc(sizeof(*log), GFP_KERNEL);
	if (!log)
		return -ENOMEM;
	log->rdev = rdev;

J
Jens Axboe 已提交
2802
	log->need_cache_flush = test_bit(QUEUE_FLAG_WC, &q->queue_flags) != 0;
2803

S
Shaohua Li 已提交
2804 2805
	log->uuid_checksum = crc32c_le(~0, rdev->mddev->uuid,
				       sizeof(rdev->mddev->uuid));
S
Shaohua Li 已提交
2806 2807 2808 2809 2810

	mutex_init(&log->io_mutex);

	spin_lock_init(&log->io_list_lock);
	INIT_LIST_HEAD(&log->running_ios);
S
Shaohua Li 已提交
2811
	INIT_LIST_HEAD(&log->io_end_ios);
2812
	INIT_LIST_HEAD(&log->flushing_ios);
2813
	INIT_LIST_HEAD(&log->finished_ios);
2814
	bio_init(&log->flush_bio, NULL, 0);
S
Shaohua Li 已提交
2815 2816 2817 2818 2819

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

2820 2821 2822 2823
	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
2824 2825 2826 2827
	log->bs = bioset_create(R5L_POOL_SIZE, 0);
	if (!log->bs)
		goto io_bs;

2828 2829 2830 2831
	log->meta_pool = mempool_create_page_pool(R5L_POOL_SIZE, 0);
	if (!log->meta_pool)
		goto out_mempool;

2832 2833 2834
	spin_lock_init(&log->tree_lock);
	INIT_RADIX_TREE(&log->big_stripe_tree, GFP_NOWAIT | __GFP_NOWARN);

S
Shaohua Li 已提交
2835 2836 2837 2838
	log->reclaim_thread = md_register_thread(r5l_reclaim_thread,
						 log->rdev->mddev, "reclaim");
	if (!log->reclaim_thread)
		goto reclaim_thread;
2839 2840
	log->reclaim_thread->timeout = R5C_RECLAIM_WAKEUP_INTERVAL;

2841
	init_waitqueue_head(&log->iounit_wait);
S
Shaohua Li 已提交
2842

2843 2844
	INIT_LIST_HEAD(&log->no_mem_stripes);

S
Shaohua Li 已提交
2845 2846 2847
	INIT_LIST_HEAD(&log->no_space_stripes);
	spin_lock_init(&log->no_space_stripes_lock);

S
Song Liu 已提交
2848
	INIT_WORK(&log->deferred_io_work, r5l_submit_io_async);
2849
	INIT_WORK(&log->disable_writeback_work, r5c_disable_writeback_async);
S
Song Liu 已提交
2850

2851
	log->r5c_journal_mode = R5C_JOURNAL_MODE_WRITE_THROUGH;
2852 2853 2854
	INIT_LIST_HEAD(&log->stripe_in_journal_list);
	spin_lock_init(&log->stripe_in_journal_lock);
	atomic_set(&log->stripe_in_journal_count, 0);
2855

2856 2857
	rcu_assign_pointer(conf->log, log);

S
Shaohua Li 已提交
2858 2859 2860
	if (r5l_load_log(log))
		goto error;

2861
	set_bit(MD_HAS_JOURNAL, &conf->mddev->flags);
S
Shaohua Li 已提交
2862
	return 0;
2863

S
Shaohua Li 已提交
2864
error:
2865
	rcu_assign_pointer(conf->log, NULL);
S
Shaohua Li 已提交
2866 2867
	md_unregister_thread(&log->reclaim_thread);
reclaim_thread:
2868 2869
	mempool_destroy(log->meta_pool);
out_mempool:
C
Christoph Hellwig 已提交
2870 2871
	bioset_free(log->bs);
io_bs:
2872 2873
	mempool_destroy(log->io_pool);
io_pool:
S
Shaohua Li 已提交
2874 2875 2876 2877 2878 2879 2880 2881
	kmem_cache_destroy(log->io_kc);
io_kc:
	kfree(log);
	return -EINVAL;
}

void r5l_exit_log(struct r5l_log *log)
{
2882
	flush_work(&log->disable_writeback_work);
S
Shaohua Li 已提交
2883
	md_unregister_thread(&log->reclaim_thread);
2884
	mempool_destroy(log->meta_pool);
C
Christoph Hellwig 已提交
2885
	bioset_free(log->bs);
2886
	mempool_destroy(log->io_pool);
S
Shaohua Li 已提交
2887 2888 2889
	kmem_cache_destroy(log->io_kc);
	kfree(log);
}