super.c 43.2 KB
Newer Older
1 2 3 4 5 6 7 8 9 10
// SPDX-License-Identifier: GPL-2.0
/*
 * bcachefs setup/teardown code, and some metadata io - read a superblock and
 * figure out what to do with it.
 *
 * Copyright 2010, 2011 Kent Overstreet <kent.overstreet@gmail.com>
 * Copyright 2012 Google, Inc.
 */

#include "bcachefs.h"
11 12
#include "alloc_background.h"
#include "alloc_foreground.h"
13
#include "bkey_sort.h"
14 15 16 17 18 19 20 21 22 23
#include "btree_cache.h"
#include "btree_gc.h"
#include "btree_update_interior.h"
#include "btree_io.h"
#include "chardev.h"
#include "checksum.h"
#include "clock.h"
#include "compress.h"
#include "debug.h"
#include "disk_groups.h"
24
#include "ec.h"
25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176
#include "error.h"
#include "fs.h"
#include "fs-io.h"
#include "fsck.h"
#include "inode.h"
#include "io.h"
#include "journal.h"
#include "journal_reclaim.h"
#include "move.h"
#include "migrate.h"
#include "movinggc.h"
#include "quota.h"
#include "rebalance.h"
#include "recovery.h"
#include "replicas.h"
#include "super.h"
#include "super-io.h"
#include "sysfs.h"
#include "trace.h"

#include <linux/backing-dev.h>
#include <linux/blkdev.h>
#include <linux/debugfs.h>
#include <linux/device.h>
#include <linux/idr.h>
#include <linux/kthread.h>
#include <linux/module.h>
#include <linux/percpu.h>
#include <linux/random.h>
#include <linux/sysfs.h>
#include <crypto/hash.h>

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Kent Overstreet <kent.overstreet@gmail.com>");

#define KTYPE(type)							\
static const struct attribute_group type ## _group = {			\
	.attrs = type ## _files						\
};									\
									\
static const struct attribute_group *type ## _groups[] = {		\
	&type ## _group,						\
	NULL								\
};									\
									\
static const struct kobj_type type ## _ktype = {			\
	.release	= type ## _release,				\
	.sysfs_ops	= &type ## _sysfs_ops,				\
	.default_groups = type ## _groups				\
}

static void bch2_fs_release(struct kobject *);
static void bch2_dev_release(struct kobject *);

static void bch2_fs_internal_release(struct kobject *k)
{
}

static void bch2_fs_opts_dir_release(struct kobject *k)
{
}

static void bch2_fs_time_stats_release(struct kobject *k)
{
}

KTYPE(bch2_fs);
KTYPE(bch2_fs_internal);
KTYPE(bch2_fs_opts_dir);
KTYPE(bch2_fs_time_stats);
KTYPE(bch2_dev);

static struct kset *bcachefs_kset;
static LIST_HEAD(bch_fs_list);
static DEFINE_MUTEX(bch_fs_list_lock);

static DECLARE_WAIT_QUEUE_HEAD(bch_read_only_wait);

static void bch2_dev_free(struct bch_dev *);
static int bch2_dev_alloc(struct bch_fs *, unsigned);
static int bch2_dev_sysfs_online(struct bch_fs *, struct bch_dev *);
static void __bch2_dev_read_only(struct bch_fs *, struct bch_dev *);

struct bch_fs *bch2_dev_to_fs(dev_t dev)
{
	struct bch_fs *c;
	struct bch_dev *ca;
	unsigned i;

	mutex_lock(&bch_fs_list_lock);
	rcu_read_lock();

	list_for_each_entry(c, &bch_fs_list, list)
		for_each_member_device_rcu(ca, c, i, NULL)
			if (ca->disk_sb.bdev->bd_dev == dev) {
				closure_get(&c->cl);
				goto found;
			}
	c = NULL;
found:
	rcu_read_unlock();
	mutex_unlock(&bch_fs_list_lock);

	return c;
}

static struct bch_fs *__bch2_uuid_to_fs(__uuid_t uuid)
{
	struct bch_fs *c;

	lockdep_assert_held(&bch_fs_list_lock);

	list_for_each_entry(c, &bch_fs_list, list)
		if (!memcmp(&c->disk_sb.sb->uuid, &uuid, sizeof(uuid)))
			return c;

	return NULL;
}

struct bch_fs *bch2_uuid_to_fs(__uuid_t uuid)
{
	struct bch_fs *c;

	mutex_lock(&bch_fs_list_lock);
	c = __bch2_uuid_to_fs(uuid);
	if (c)
		closure_get(&c->cl);
	mutex_unlock(&bch_fs_list_lock);

	return c;
}

/* Filesystem RO/RW: */

/*
 * For startup/shutdown of RW stuff, the dependencies are:
 *
 * - foreground writes depend on copygc and rebalance (to free up space)
 *
 * - copygc and rebalance depend on mark and sweep gc (they actually probably
 *   don't because they either reserve ahead of time or don't block if
 *   allocations fail, but allocations can require mark and sweep gc to run
 *   because of generation number wraparound)
 *
 * - all of the above depends on the allocator threads
 *
 * - allocator depends on the journal (when it rewrites prios and gens)
 */

static void __bch2_fs_read_only(struct bch_fs *c)
{
	struct bch_dev *ca;
177
	bool wrote;
178
	unsigned i;
179
	int ret;
180 181 182 183 184 185 186 187 188 189 190 191 192 193

	bch2_rebalance_stop(c);

	for_each_member_device(ca, c, i)
		bch2_copygc_stop(ca);

	bch2_gc_thread_stop(c);

	/*
	 * Flush journal before stopping allocators, because flushing journal
	 * blacklist entries involves allocating new btree nodes:
	 */
	bch2_journal_flush_all_pins(&c->journal);

194 195 196
	if (!test_bit(BCH_FS_ALLOCATOR_RUNNING, &c->flags))
		goto allocator_not_running;

197 198 199 200 201 202 203
	do {
		ret = bch2_alloc_write(c, false, &wrote);
		if (ret) {
			bch2_fs_inconsistent(c, "error writing out alloc info %i", ret);
			break;
		}

204 205 206 207 208 209
		ret = bch2_stripes_write(c, &wrote);
		if (ret) {
			bch2_fs_inconsistent(c, "error writing out stripes");
			break;
		}

210 211 212 213 214 215 216 217 218 219 220 221 222 223 224
		for_each_member_device(ca, c, i)
			bch2_dev_allocator_quiesce(c, ca);

		bch2_journal_flush_all_pins(&c->journal);

		/*
		 * We need to explicitly wait on btree interior updates to complete
		 * before stopping the journal, flushing all journal pins isn't
		 * sufficient, because in the BTREE_INTERIOR_UPDATING_ROOT case btree
		 * interior updates have to drop their journal pin before they're
		 * fully complete:
		 */
		closure_wait_event(&c->btree_interior_update_wait,
				   !bch2_btree_interior_updates_nr_pending(c));
	} while (wrote);
225
allocator_not_running:
226 227 228
	for_each_member_device(ca, c, i)
		bch2_dev_allocator_stop(ca);

229 230
	clear_bit(BCH_FS_ALLOCATOR_RUNNING, &c->flags);

231 232
	bch2_fs_journal_stop(&c->journal);

233 234
	/* XXX: mark super that alloc info is persistent */

235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260
	/*
	 * the journal kicks off btree writes via reclaim - wait for in flight
	 * writes after stopping journal:
	 */
	if (test_bit(BCH_FS_EMERGENCY_RO, &c->flags))
		bch2_btree_flush_all_writes(c);
	else
		bch2_btree_verify_flushed(c);

	/*
	 * After stopping journal:
	 */
	for_each_member_device(ca, c, i)
		bch2_dev_allocator_remove(c, ca);
}

static void bch2_writes_disabled(struct percpu_ref *writes)
{
	struct bch_fs *c = container_of(writes, struct bch_fs, writes);

	set_bit(BCH_FS_WRITE_DISABLE_COMPLETE, &c->flags);
	wake_up(&bch_read_only_wait);
}

void bch2_fs_read_only(struct bch_fs *c)
{
Kent Overstreet's avatar
Kent Overstreet committed
261 262
	if (!test_bit(BCH_FS_RW, &c->flags)) {
		cancel_delayed_work_sync(&c->journal.reclaim_work);
263
		return;
Kent Overstreet's avatar
Kent Overstreet committed
264
	}
265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303

	BUG_ON(test_bit(BCH_FS_WRITE_DISABLE_COMPLETE, &c->flags));

	/*
	 * Block new foreground-end write operations from starting - any new
	 * writes will return -EROFS:
	 *
	 * (This is really blocking new _allocations_, writes to previously
	 * allocated space can still happen until stopping the allocator in
	 * bch2_dev_allocator_stop()).
	 */
	percpu_ref_kill(&c->writes);

	cancel_delayed_work(&c->pd_controllers_update);

	/*
	 * If we're not doing an emergency shutdown, we want to wait on
	 * outstanding writes to complete so they don't see spurious errors due
	 * to shutting down the allocator:
	 *
	 * If we are doing an emergency shutdown outstanding writes may
	 * hang until we shutdown the allocator so we don't want to wait
	 * on outstanding writes before shutting everything down - but
	 * we do need to wait on them before returning and signalling
	 * that going RO is complete:
	 */
	wait_event(bch_read_only_wait,
		   test_bit(BCH_FS_WRITE_DISABLE_COMPLETE, &c->flags) ||
		   test_bit(BCH_FS_EMERGENCY_RO, &c->flags));

	__bch2_fs_read_only(c);

	wait_event(bch_read_only_wait,
		   test_bit(BCH_FS_WRITE_DISABLE_COMPLETE, &c->flags));

	clear_bit(BCH_FS_WRITE_DISABLE_COMPLETE, &c->flags);

	if (!bch2_journal_error(&c->journal) &&
	    !test_bit(BCH_FS_ERROR, &c->flags) &&
304 305
	    !test_bit(BCH_FS_EMERGENCY_RO, &c->flags) &&
	    test_bit(BCH_FS_STARTED, &c->flags))
Kent Overstreet's avatar
Kent Overstreet committed
306
		bch2_fs_mark_clean(c);
307

Kent Overstreet's avatar
Kent Overstreet committed
308
	clear_bit(BCH_FS_RW, &c->flags);
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
}

static void bch2_fs_read_only_work(struct work_struct *work)
{
	struct bch_fs *c =
		container_of(work, struct bch_fs, read_only_work);

	mutex_lock(&c->state_lock);
	bch2_fs_read_only(c);
	mutex_unlock(&c->state_lock);
}

static void bch2_fs_read_only_async(struct bch_fs *c)
{
	queue_work(system_long_wq, &c->read_only_work);
}

bool bch2_fs_emergency_read_only(struct bch_fs *c)
{
	bool ret = !test_and_set_bit(BCH_FS_EMERGENCY_RO, &c->flags);

	bch2_fs_read_only_async(c);
	bch2_journal_halt(&c->journal);

	wake_up(&bch_read_only_wait);
	return ret;
}

Kent Overstreet's avatar
Kent Overstreet committed
337
static int bch2_fs_read_write_late(struct bch_fs *c)
338 339 340
{
	struct bch_dev *ca;
	unsigned i;
Kent Overstreet's avatar
Kent Overstreet committed
341
	int ret;
342

Kent Overstreet's avatar
Kent Overstreet committed
343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367
	ret = bch2_gc_thread_start(c);
	if (ret) {
		bch_err(c, "error starting gc thread");
		return ret;
	}

	for_each_rw_member(ca, c, i) {
		ret = bch2_copygc_start(c, ca);
		if (ret) {
			bch_err(c, "error starting copygc threads");
			percpu_ref_put(&ca->io_ref);
			return ret;
		}
	}

	ret = bch2_rebalance_start(c);
	if (ret) {
		bch_err(c, "error starting rebalance thread");
		return ret;
	}

	schedule_delayed_work(&c->pd_controllers_update, 5 * HZ);

	return 0;
}
368

Kent Overstreet's avatar
Kent Overstreet committed
369 370 371 372 373 374 375 376 377 378 379 380
static int __bch2_fs_read_write(struct bch_fs *c, bool early)
{
	struct bch_dev *ca;
	unsigned i;
	int ret;

	if (test_bit(BCH_FS_RW, &c->flags))
		return 0;

	ret = bch2_fs_mark_dirty(c);
	if (ret)
		goto err;
381 382 383 384 385

	for_each_rw_member(ca, c, i)
		bch2_dev_allocator_add(c, ca);
	bch2_recalc_capacity(c);

Kent Overstreet's avatar
Kent Overstreet committed
386 387 388 389
	if (!test_bit(BCH_FS_ALLOCATOR_STARTED, &c->flags)) {
		ret = bch2_fs_allocator_start(c);
		if (ret) {
			bch_err(c, "error initializing allocator");
390 391 392
			goto err;
		}

Kent Overstreet's avatar
Kent Overstreet committed
393 394
		set_bit(BCH_FS_ALLOCATOR_STARTED, &c->flags);
	}
395

Kent Overstreet's avatar
Kent Overstreet committed
396 397 398 399
	for_each_rw_member(ca, c, i) {
		ret = bch2_dev_allocator_start(ca);
		if (ret) {
			bch_err(c, "error starting allocator threads");
400 401 402
			percpu_ref_put(&ca->io_ref);
			goto err;
		}
Kent Overstreet's avatar
Kent Overstreet committed
403
	}
404

Kent Overstreet's avatar
Kent Overstreet committed
405
	set_bit(BCH_FS_ALLOCATOR_RUNNING, &c->flags);
406

Kent Overstreet's avatar
Kent Overstreet committed
407 408 409 410 411
	if (!early) {
		ret = bch2_fs_read_write_late(c);
		if (ret)
			goto err;
	}
412

Kent Overstreet's avatar
Kent Overstreet committed
413 414
	percpu_ref_reinit(&c->writes);
	set_bit(BCH_FS_RW, &c->flags);
415

Kent Overstreet's avatar
Kent Overstreet committed
416 417 418
	queue_delayed_work(c->journal_reclaim_wq,
			   &c->journal.reclaim_work, 0);
	return 0;
419 420
err:
	__bch2_fs_read_only(c);
Kent Overstreet's avatar
Kent Overstreet committed
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436
	return ret;
}

int bch2_fs_read_write(struct bch_fs *c)
{
	return __bch2_fs_read_write(c, false);
}

int bch2_fs_read_write_early(struct bch_fs *c)
{
	lockdep_assert_held(&c->state_lock);

	if (c->opts.read_only)
		return -EROFS;

	return __bch2_fs_read_write(c, true);
437 438 439 440 441 442 443 444 445 446 447 448 449
}

/* Filesystem startup/shutdown: */

static void bch2_fs_free(struct bch_fs *c)
{
	unsigned i;

	for (i = 0; i < BCH_TIME_STAT_NR; i++)
		bch2_time_stats_exit(&c->times[i]);

	bch2_fs_quota_exit(c);
	bch2_fs_fsio_exit(c);
450
	bch2_fs_ec_exit(c);
451 452 453 454 455 456 457
	bch2_fs_encryption_exit(c);
	bch2_fs_io_exit(c);
	bch2_fs_btree_cache_exit(c);
	bch2_fs_journal_exit(&c->journal);
	bch2_io_clock_exit(&c->io_clock[WRITE]);
	bch2_io_clock_exit(&c->io_clock[READ]);
	bch2_fs_compress_exit(c);
458
	percpu_free_rwsem(&c->mark_lock);
459
	kfree(c->usage_scratch);
460
	free_percpu(c->usage[0]);
461
	free_percpu(c->pcpu);
462
	mempool_exit(&c->btree_iters_pool);
463 464 465 466 467 468
	mempool_exit(&c->btree_bounce_pool);
	bioset_exit(&c->btree_bio);
	mempool_exit(&c->btree_interior_update_pool);
	mempool_exit(&c->btree_reserve_pool);
	mempool_exit(&c->fill_iter);
	percpu_ref_exit(&c->writes);
469 470
	kfree(c->replicas.entries);
	kfree(c->replicas_gc.entries);
471 472
	kfree(rcu_dereference_protected(c->disk_groups, 1));

473 474
	if (c->journal_reclaim_wq)
		destroy_workqueue(c->journal_reclaim_wq);
475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497
	if (c->copygc_wq)
		destroy_workqueue(c->copygc_wq);
	if (c->wq)
		destroy_workqueue(c->wq);

	free_pages((unsigned long) c->disk_sb.sb,
		   c->disk_sb.page_order);
	kvpfree(c, sizeof(*c));
	module_put(THIS_MODULE);
}

static void bch2_fs_release(struct kobject *kobj)
{
	struct bch_fs *c = container_of(kobj, struct bch_fs, kobj);

	bch2_fs_free(c);
}

void bch2_fs_stop(struct bch_fs *c)
{
	struct bch_dev *ca;
	unsigned i;

498 499
	bch_verbose(c, "shutting down");

500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
	for_each_member_device(ca, c, i)
		if (ca->kobj.state_in_sysfs &&
		    ca->disk_sb.bdev)
			sysfs_remove_link(bdev_kobj(ca->disk_sb.bdev), "bcachefs");

	if (c->kobj.state_in_sysfs)
		kobject_del(&c->kobj);

	bch2_fs_debug_exit(c);
	bch2_fs_chardev_exit(c);

	kobject_put(&c->time_stats);
	kobject_put(&c->opts_dir);
	kobject_put(&c->internal);

	mutex_lock(&bch_fs_list_lock);
	list_del(&c->list);
	mutex_unlock(&bch_fs_list_lock);

	closure_sync(&c->cl);
	closure_debug_destroy(&c->cl);

	mutex_lock(&c->state_lock);
	bch2_fs_read_only(c);
	mutex_unlock(&c->state_lock);

	/* btree prefetch might have kicked off reads in the background: */
	bch2_btree_flush_all_reads(c);

	for_each_member_device(ca, c, i)
		cancel_work_sync(&ca->io_error_work);

	cancel_work_sync(&c->btree_write_error_work);
	cancel_delayed_work_sync(&c->pd_controllers_update);
	cancel_work_sync(&c->read_only_work);

	for (i = 0; i < c->sb.nr_devices; i++)
		if (c->devs[i])
			bch2_dev_free(rcu_dereference_protected(c->devs[i], 1));

540 541
	bch_verbose(c, "shutdown complete");

542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590
	kobject_put(&c->kobj);
}

static const char *bch2_fs_online(struct bch_fs *c)
{
	struct bch_dev *ca;
	const char *err = NULL;
	unsigned i;
	int ret;

	lockdep_assert_held(&bch_fs_list_lock);

	if (!list_empty(&c->list))
		return NULL;

	if (__bch2_uuid_to_fs(c->sb.uuid))
		return "filesystem UUID already open";

	ret = bch2_fs_chardev_init(c);
	if (ret)
		return "error creating character device";

	bch2_fs_debug_init(c);

	if (kobject_add(&c->kobj, NULL, "%pU", c->sb.user_uuid.b) ||
	    kobject_add(&c->internal, &c->kobj, "internal") ||
	    kobject_add(&c->opts_dir, &c->kobj, "options") ||
	    kobject_add(&c->time_stats, &c->kobj, "time_stats") ||
	    bch2_opts_create_sysfs_files(&c->opts_dir))
		return "error creating sysfs objects";

	mutex_lock(&c->state_lock);

	err = "error creating sysfs objects";
	__for_each_member_device(ca, c, i, NULL)
		if (bch2_dev_sysfs_online(c, ca))
			goto err;

	list_add(&c->list, &bch_fs_list);
	err = NULL;
err:
	mutex_unlock(&c->state_lock);
	return err;
}

static struct bch_fs *bch2_fs_alloc(struct bch_sb *sb, struct bch_opts opts)
{
	struct bch_sb_field_members *mi;
	struct bch_fs *c;
591
	unsigned i, iter_size;
592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615
	const char *err;

	pr_verbose_init(opts, "");

	c = kvpmalloc(sizeof(struct bch_fs), GFP_KERNEL|__GFP_ZERO);
	if (!c)
		goto out;

	__module_get(THIS_MODULE);

	c->minor		= -1;
	c->disk_sb.fs_sb	= true;

	mutex_init(&c->state_lock);
	mutex_init(&c->sb_lock);
	mutex_init(&c->replicas_gc_lock);
	mutex_init(&c->btree_root_lock);
	INIT_WORK(&c->read_only_work, bch2_fs_read_only_work);

	init_rwsem(&c->gc_lock);

	for (i = 0; i < BCH_TIME_STAT_NR; i++)
		bch2_time_stats_init(&c->times[i]);

616 617
	bch2_fs_allocator_background_init(c);
	bch2_fs_allocator_foreground_init(c);
618 619 620 621 622 623 624 625 626
	bch2_fs_rebalance_init(c);
	bch2_fs_quota_init(c);

	INIT_LIST_HEAD(&c->list);

	INIT_LIST_HEAD(&c->btree_interior_update_list);
	mutex_init(&c->btree_reserve_cache_lock);
	mutex_init(&c->btree_interior_update_lock);

627 628
	mutex_init(&c->usage_scratch_lock);

629 630 631 632 633 634 635 636 637
	mutex_init(&c->bio_bounce_pages_lock);

	bio_list_init(&c->btree_write_error_list);
	spin_lock_init(&c->btree_write_error_lock);
	INIT_WORK(&c->btree_write_error_work, bch2_btree_write_error_work);

	INIT_LIST_HEAD(&c->fsck_errors);
	mutex_init(&c->fsck_error_lock);

638 639
	INIT_LIST_HEAD(&c->ec_new_stripe_list);
	mutex_init(&c->ec_new_stripe_lock);
640
	mutex_init(&c->ec_stripe_create_lock);
641 642
	spin_lock_init(&c->ec_stripes_heap_lock);

643 644 645 646 647 648 649 650
	seqcount_init(&c->gc_pos_lock);

	c->copy_gc_enabled		= 1;
	c->rebalance.enabled		= 1;
	c->promote_whole_extents	= true;

	c->journal.write_time	= &c->times[BCH_TIME_journal_write];
	c->journal.delay_time	= &c->times[BCH_TIME_journal_delay];
651
	c->journal.blocked_time	= &c->times[BCH_TIME_blocked_journal];
652 653 654 655
	c->journal.flush_seq_time = &c->times[BCH_TIME_journal_flush_seq];

	bch2_fs_btree_cache_init_early(&c->btree_cache);

656 657 658
	if (percpu_init_rwsem(&c->mark_lock))
		goto err;

659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690
	mutex_lock(&c->sb_lock);

	if (bch2_sb_to_fs(c, sb)) {
		mutex_unlock(&c->sb_lock);
		goto err;
	}

	mutex_unlock(&c->sb_lock);

	scnprintf(c->name, sizeof(c->name), "%pU", &c->sb.user_uuid);

	c->opts = bch2_opts_default;
	bch2_opts_apply(&c->opts, bch2_opts_from_sb(sb));
	bch2_opts_apply(&c->opts, opts);

	c->block_bits		= ilog2(c->opts.block_size);
	c->btree_foreground_merge_threshold = BTREE_FOREGROUND_MERGE_THRESHOLD(c);

	c->opts.nochanges	|= c->opts.noreplay;
	c->opts.read_only	|= c->opts.nochanges;

	if (bch2_fs_init_fault("fs_alloc"))
		goto err;

	iter_size = sizeof(struct btree_node_iter_large) +
		(btree_blocks(c) + 1) * 2 *
		sizeof(struct btree_node_iter_set);

	if (!(c->wq = alloc_workqueue("bcachefs",
				WQ_FREEZABLE|WQ_MEM_RECLAIM|WQ_HIGHPRI, 1)) ||
	    !(c->copygc_wq = alloc_workqueue("bcache_copygc",
				WQ_FREEZABLE|WQ_MEM_RECLAIM|WQ_HIGHPRI, 1)) ||
691 692
	    !(c->journal_reclaim_wq = alloc_workqueue("bcache_journal",
				WQ_FREEZABLE|WQ_MEM_RECLAIM|WQ_HIGHPRI, 1)) ||
Kent Overstreet's avatar
Kent Overstreet committed
693 694
	    percpu_ref_init(&c->writes, bch2_writes_disabled,
			    PERCPU_REF_INIT_DEAD, GFP_KERNEL) ||
695 696 697 698 699 700 701 702 703
	    mempool_init_kmalloc_pool(&c->btree_reserve_pool, 1,
				      sizeof(struct btree_reserve)) ||
	    mempool_init_kmalloc_pool(&c->btree_interior_update_pool, 1,
				      sizeof(struct btree_update)) ||
	    mempool_init_kmalloc_pool(&c->fill_iter, 1, iter_size) ||
	    bioset_init(&c->btree_bio, 1,
			max(offsetof(struct btree_read_bio, bio),
			    offsetof(struct btree_write_bio, wbio.bio)),
			BIOSET_NEED_BVECS) ||
704
	    !(c->pcpu = alloc_percpu(struct bch_fs_pcpu)) ||
705 706
	    mempool_init_kvpmalloc_pool(&c->btree_bounce_pool, 1,
					btree_bytes(c)) ||
707
	    mempool_init_kmalloc_pool(&c->btree_iters_pool, 1,
708 709 710
			sizeof(struct btree_iter) * BTREE_ITER_MAX +
			sizeof(struct btree_insert_entry) *
			(BTREE_ITER_MAX + 4)) ||
711 712 713
	    bch2_io_clock_init(&c->io_clock[READ]) ||
	    bch2_io_clock_init(&c->io_clock[WRITE]) ||
	    bch2_fs_journal_init(&c->journal) ||
714
	    bch2_fs_replicas_init(c) ||
715 716 717 718
	    bch2_fs_btree_cache_init(c) ||
	    bch2_fs_io_init(c) ||
	    bch2_fs_encryption_init(c) ||
	    bch2_fs_compress_init(c) ||
719
	    bch2_fs_ec_init(c) ||
720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761
	    bch2_fs_fsio_init(c))
		goto err;

	mi = bch2_sb_get_members(c->disk_sb.sb);
	for (i = 0; i < c->sb.nr_devices; i++)
		if (bch2_dev_exists(c->disk_sb.sb, mi, i) &&
		    bch2_dev_alloc(c, i))
			goto err;

	/*
	 * Now that all allocations have succeeded, init various refcounty
	 * things that let us shutdown:
	 */
	closure_init(&c->cl, NULL);

	c->kobj.kset = bcachefs_kset;
	kobject_init(&c->kobj, &bch2_fs_ktype);
	kobject_init(&c->internal, &bch2_fs_internal_ktype);
	kobject_init(&c->opts_dir, &bch2_fs_opts_dir_ktype);
	kobject_init(&c->time_stats, &bch2_fs_time_stats_ktype);

	mutex_lock(&bch_fs_list_lock);
	err = bch2_fs_online(c);
	mutex_unlock(&bch_fs_list_lock);
	if (err) {
		bch_err(c, "bch2_fs_online() error: %s", err);
		goto err;
	}
out:
	pr_verbose_init(opts, "ret %i", c ? 0 : -ENOMEM);
	return c;
err:
	bch2_fs_free(c);
	c = NULL;
	goto out;
}

const char *bch2_fs_start(struct bch_fs *c)
{
	const char *err = "cannot allocate memory";
	struct bch_sb_field_members *mi;
	struct bch_dev *ca;
762
	time64_t now = ktime_get_real_seconds();
763 764 765 766 767
	unsigned i;
	int ret = -EINVAL;

	mutex_lock(&c->state_lock);

Kent Overstreet's avatar
Kent Overstreet committed
768
	BUG_ON(test_bit(BCH_FS_STARTED, &c->flags));
769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789

	mutex_lock(&c->sb_lock);

	for_each_online_member(ca, c, i)
		bch2_sb_from_fs(c, ca);

	mi = bch2_sb_get_members(c->disk_sb.sb);
	for_each_online_member(ca, c, i)
		mi->members[ca->dev_idx].last_mount = cpu_to_le64(now);

	mutex_unlock(&c->sb_lock);

	for_each_rw_member(ca, c, i)
		bch2_dev_allocator_add(c, ca);
	bch2_recalc_capacity(c);

	ret = BCH_SB_INITIALIZED(c->disk_sb.sb)
		? bch2_fs_recovery(c)
		: bch2_fs_initialize(c);
	if (ret)
		goto err;
790 791 792 793

	ret = bch2_opts_check_may_set(c);
	if (ret)
		goto err;
794 795 796 797 798 799 800 801

	err = "dynamic fault";
	if (bch2_fs_init_fault("fs_start"))
		goto err;

	if (c->opts.read_only) {
		bch2_fs_read_only(c);
	} else {
Kent Overstreet's avatar
Kent Overstreet committed
802 803 804 805
		if (!test_bit(BCH_FS_RW, &c->flags)
		    ? bch2_fs_read_write(c)
		    : bch2_fs_read_write_late(c)) {
			err = "error going read write";
806
			goto err;
Kent Overstreet's avatar
Kent Overstreet committed
807
		}
808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 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 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129
	}

	set_bit(BCH_FS_STARTED, &c->flags);

	err = NULL;
out:
	mutex_unlock(&c->state_lock);
	return err;
err:
	switch (ret) {
	case BCH_FSCK_ERRORS_NOT_FIXED:
		bch_err(c, "filesystem contains errors: please report this to the developers");
		pr_cont("mount with -o fix_errors to repair\n");
		err = "fsck error";
		break;
	case BCH_FSCK_REPAIR_UNIMPLEMENTED:
		bch_err(c, "filesystem contains errors: please report this to the developers");
		pr_cont("repair unimplemented: inform the developers so that it can be added\n");
		err = "fsck error";
		break;
	case BCH_FSCK_REPAIR_IMPOSSIBLE:
		bch_err(c, "filesystem contains errors, but repair impossible");
		err = "fsck error";
		break;
	case BCH_FSCK_UNKNOWN_VERSION:
		err = "unknown metadata version";;
		break;
	case -ENOMEM:
		err = "cannot allocate memory";
		break;
	case -EIO:
		err = "IO error";
		break;
	}

	BUG_ON(!err);
	set_bit(BCH_FS_ERROR, &c->flags);
	goto out;
}

static const char *bch2_dev_may_add(struct bch_sb *sb, struct bch_fs *c)
{
	struct bch_sb_field_members *sb_mi;

	sb_mi = bch2_sb_get_members(sb);
	if (!sb_mi)
		return "Invalid superblock: member info area missing";

	if (le16_to_cpu(sb->block_size) != c->opts.block_size)
		return "mismatched block size";

	if (le16_to_cpu(sb_mi->members[sb->dev_idx].bucket_size) <
	    BCH_SB_BTREE_NODE_SIZE(c->disk_sb.sb))
		return "new cache bucket size is too small";

	return NULL;
}

static const char *bch2_dev_in_fs(struct bch_sb *fs, struct bch_sb *sb)
{
	struct bch_sb *newest =
		le64_to_cpu(fs->seq) > le64_to_cpu(sb->seq) ? fs : sb;
	struct bch_sb_field_members *mi = bch2_sb_get_members(newest);

	if (!uuid_equal(&fs->uuid, &sb->uuid))
		return "device not a member of filesystem";

	if (!bch2_dev_exists(newest, mi, sb->dev_idx))
		return "device has been removed";

	if (fs->block_size != sb->block_size)
		return "mismatched block size";

	return NULL;
}

/* Device startup/shutdown: */

static void bch2_dev_release(struct kobject *kobj)
{
	struct bch_dev *ca = container_of(kobj, struct bch_dev, kobj);

	kfree(ca);
}

static void bch2_dev_free(struct bch_dev *ca)
{
	cancel_work_sync(&ca->io_error_work);

	if (ca->kobj.state_in_sysfs &&
	    ca->disk_sb.bdev)
		sysfs_remove_link(bdev_kobj(ca->disk_sb.bdev), "bcachefs");

	if (ca->kobj.state_in_sysfs)
		kobject_del(&ca->kobj);

	bch2_free_super(&ca->disk_sb);
	bch2_dev_journal_exit(ca);

	free_percpu(ca->io_done);
	bioset_exit(&ca->replica_set);
	bch2_dev_buckets_free(ca);

	bch2_time_stats_exit(&ca->io_latency[WRITE]);
	bch2_time_stats_exit(&ca->io_latency[READ]);

	percpu_ref_exit(&ca->io_ref);
	percpu_ref_exit(&ca->ref);
	kobject_put(&ca->kobj);
}

static void __bch2_dev_offline(struct bch_fs *c, struct bch_dev *ca)
{

	lockdep_assert_held(&c->state_lock);

	if (percpu_ref_is_zero(&ca->io_ref))
		return;

	__bch2_dev_read_only(c, ca);

	reinit_completion(&ca->io_ref_completion);
	percpu_ref_kill(&ca->io_ref);
	wait_for_completion(&ca->io_ref_completion);

	if (ca->kobj.state_in_sysfs) {
		sysfs_remove_link(bdev_kobj(ca->disk_sb.bdev), "bcachefs");
		sysfs_remove_link(&ca->kobj, "block");
	}

	bch2_free_super(&ca->disk_sb);
	bch2_dev_journal_exit(ca);
}

static void bch2_dev_ref_complete(struct percpu_ref *ref)
{
	struct bch_dev *ca = container_of(ref, struct bch_dev, ref);

	complete(&ca->ref_completion);
}

static void bch2_dev_io_ref_complete(struct percpu_ref *ref)
{
	struct bch_dev *ca = container_of(ref, struct bch_dev, io_ref);

	complete(&ca->io_ref_completion);
}

static int bch2_dev_sysfs_online(struct bch_fs *c, struct bch_dev *ca)
{
	int ret;

	if (!c->kobj.state_in_sysfs)
		return 0;

	if (!ca->kobj.state_in_sysfs) {
		ret = kobject_add(&ca->kobj, &c->kobj,
				  "dev-%u", ca->dev_idx);
		if (ret)
			return ret;
	}

	if (ca->disk_sb.bdev) {
		struct kobject *block = bdev_kobj(ca->disk_sb.bdev);

		ret = sysfs_create_link(block, &ca->kobj, "bcachefs");
		if (ret)
			return ret;

		ret = sysfs_create_link(&ca->kobj, block, "block");
		if (ret)
			return ret;
	}

	return 0;
}

static struct bch_dev *__bch2_dev_alloc(struct bch_fs *c,
					struct bch_member *member)
{
	struct bch_dev *ca;

	ca = kzalloc(sizeof(*ca), GFP_KERNEL);
	if (!ca)
		return NULL;

	kobject_init(&ca->kobj, &bch2_dev_ktype);
	init_completion(&ca->ref_completion);
	init_completion(&ca->io_ref_completion);

	init_rwsem(&ca->bucket_lock);

	writepoint_init(&ca->copygc_write_point, BCH_DATA_USER);

	spin_lock_init(&ca->freelist_lock);
	bch2_dev_copygc_init(ca);

	INIT_WORK(&ca->io_error_work, bch2_io_error_work);

	bch2_time_stats_init(&ca->io_latency[READ]);
	bch2_time_stats_init(&ca->io_latency[WRITE]);

	ca->mi = bch2_mi_to_cpu(member);
	ca->uuid = member->uuid;

	if (opt_defined(c->opts, discard))
		ca->mi.discard = opt_get(c->opts, discard);

	if (percpu_ref_init(&ca->ref, bch2_dev_ref_complete,
			    0, GFP_KERNEL) ||
	    percpu_ref_init(&ca->io_ref, bch2_dev_io_ref_complete,
			    PERCPU_REF_INIT_DEAD, GFP_KERNEL) ||
	    bch2_dev_buckets_alloc(c, ca) ||
	    bioset_init(&ca->replica_set, 4,
			offsetof(struct bch_write_bio, bio), 0) ||
	    !(ca->io_done	= alloc_percpu(*ca->io_done)))
		goto err;

	return ca;
err:
	bch2_dev_free(ca);
	return NULL;
}

static void bch2_dev_attach(struct bch_fs *c, struct bch_dev *ca,
			    unsigned dev_idx)
{
	ca->dev_idx = dev_idx;
	__set_bit(ca->dev_idx, ca->self.d);
	scnprintf(ca->name, sizeof(ca->name), "dev-%u", dev_idx);

	ca->fs = c;
	rcu_assign_pointer(c->devs[ca->dev_idx], ca);

	if (bch2_dev_sysfs_online(c, ca))
		pr_warn("error creating sysfs objects");
}

static int bch2_dev_alloc(struct bch_fs *c, unsigned dev_idx)
{
	struct bch_member *member =
		bch2_sb_get_members(c->disk_sb.sb)->members + dev_idx;
	struct bch_dev *ca = NULL;
	int ret = 0;

	pr_verbose_init(c->opts, "");

	if (bch2_fs_init_fault("dev_alloc"))
		goto err;

	ca = __bch2_dev_alloc(c, member);
	if (!ca)
		goto err;

	bch2_dev_attach(c, ca, dev_idx);
out:
	pr_verbose_init(c->opts, "ret %i", ret);
	return ret;
err:
	if (ca)
		bch2_dev_free(ca);
	ret = -ENOMEM;
	goto out;
}

static int __bch2_dev_attach_bdev(struct bch_dev *ca, struct bch_sb_handle *sb)
{
	unsigned ret;

	if (bch2_dev_is_online(ca)) {
		bch_err(ca, "already have device online in slot %u",
			sb->sb->dev_idx);
		return -EINVAL;
	}

	if (get_capacity(sb->bdev->bd_disk) <
	    ca->mi.bucket_size * ca->mi.nbuckets) {
		bch_err(ca, "cannot online: device too small");
		return -EINVAL;
	}

	BUG_ON(!percpu_ref_is_zero(&ca->io_ref));

	if (get_capacity(sb->bdev->bd_disk) <
	    ca->mi.bucket_size * ca->mi.nbuckets) {
		bch_err(ca, "device too small");
		return -EINVAL;
	}

	ret = bch2_dev_journal_init(ca, sb->sb);
	if (ret)
		return ret;

	/* Commit: */
	ca->disk_sb = *sb;
	memset(sb, 0, sizeof(*sb));

	percpu_ref_reinit(&ca->io_ref);

	return 0;
}

static int bch2_dev_attach_bdev(struct bch_fs *c, struct bch_sb_handle *sb)
{
	struct bch_dev *ca;
	int ret;

	lockdep_assert_held(&c->state_lock);

	if (le64_to_cpu(sb->sb->seq) >
	    le64_to_cpu(c->disk_sb.sb->seq))
		bch2_sb_to_fs(c, sb->sb);

	BUG_ON(sb->sb->dev_idx >= c->sb.nr_devices ||
	       !c->devs[sb->sb->dev_idx]);

	ca = bch_dev_locked(c, sb->sb->dev_idx);

	ret = __bch2_dev_attach_bdev(ca, sb);
	if (ret)
		return ret;

1130 1131 1132 1133 1134 1135
	if (test_bit(BCH_FS_ALLOC_READ_DONE, &c->flags) &&
	    !percpu_u64_get(&ca->usage[0]->buckets[BCH_DATA_SB])) {
		mutex_lock(&c->sb_lock);
		bch2_mark_dev_superblock(ca->fs, ca, 0);
		mutex_unlock(&c->sb_lock);
	}
1136

1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352
	bch2_dev_sysfs_online(c, ca);

	if (c->sb.nr_devices == 1)
		snprintf(c->name, sizeof(c->name), "%pg", ca->disk_sb.bdev);
	snprintf(ca->name, sizeof(ca->name), "%pg", ca->disk_sb.bdev);

	rebalance_wakeup(c);
	return 0;
}

/* Device management: */

/*
 * Note: this function is also used by the error paths - when a particular
 * device sees an error, we call it to determine whether we can just set the
 * device RO, or - if this function returns false - we'll set the whole
 * filesystem RO:
 *
 * XXX: maybe we should be more explicit about whether we're changing state
 * because we got an error or what have you?
 */
bool bch2_dev_state_allowed(struct bch_fs *c, struct bch_dev *ca,
			    enum bch_member_state new_state, int flags)
{
	struct bch_devs_mask new_online_devs;
	struct replicas_status s;
	struct bch_dev *ca2;
	int i, nr_rw = 0, required;

	lockdep_assert_held(&c->state_lock);

	switch (new_state) {
	case BCH_MEMBER_STATE_RW:
		return true;
	case BCH_MEMBER_STATE_RO:
		if (ca->mi.state != BCH_MEMBER_STATE_RW)
			return true;

		/* do we have enough devices to write to?  */
		for_each_member_device(ca2, c, i)
			if (ca2 != ca)
				nr_rw += ca2->mi.state == BCH_MEMBER_STATE_RW;

		required = max(!(flags & BCH_FORCE_IF_METADATA_DEGRADED)
			       ? c->opts.metadata_replicas
			       : c->opts.metadata_replicas_required,
			       !(flags & BCH_FORCE_IF_DATA_DEGRADED)
			       ? c->opts.data_replicas
			       : c->opts.data_replicas_required);

		return nr_rw >= required;
	case BCH_MEMBER_STATE_FAILED:
	case BCH_MEMBER_STATE_SPARE:
		if (ca->mi.state != BCH_MEMBER_STATE_RW &&
		    ca->mi.state != BCH_MEMBER_STATE_RO)
			return true;

		/* do we have enough devices to read from?  */
		new_online_devs = bch2_online_devs(c);
		__clear_bit(ca->dev_idx, new_online_devs.d);

		s = __bch2_replicas_status(c, new_online_devs);

		return bch2_have_enough_devs(s, flags);
	default:
		BUG();
	}
}

static bool bch2_fs_may_start(struct bch_fs *c)
{
	struct replicas_status s;
	struct bch_sb_field_members *mi;
	struct bch_dev *ca;
	unsigned i, flags = c->opts.degraded
		? BCH_FORCE_IF_DEGRADED
		: 0;

	if (!c->opts.degraded) {
		mutex_lock(&c->sb_lock);
		mi = bch2_sb_get_members(c->disk_sb.sb);

		for (i = 0; i < c->disk_sb.sb->nr_devices; i++) {
			if (!bch2_dev_exists(c->disk_sb.sb, mi, i))
				continue;

			ca = bch_dev_locked(c, i);

			if (!bch2_dev_is_online(ca) &&
			    (ca->mi.state == BCH_MEMBER_STATE_RW ||
			     ca->mi.state == BCH_MEMBER_STATE_RO)) {
				mutex_unlock(&c->sb_lock);
				return false;
			}
		}
		mutex_unlock(&c->sb_lock);
	}

	s = bch2_replicas_status(c);

	return bch2_have_enough_devs(s, flags);
}

static void __bch2_dev_read_only(struct bch_fs *c, struct bch_dev *ca)
{
	bch2_copygc_stop(ca);

	/*
	 * The allocator thread itself allocates btree nodes, so stop it first:
	 */
	bch2_dev_allocator_stop(ca);
	bch2_dev_allocator_remove(c, ca);
	bch2_dev_journal_stop(&c->journal, ca);
}

static const char *__bch2_dev_read_write(struct bch_fs *c, struct bch_dev *ca)
{
	lockdep_assert_held(&c->state_lock);

	BUG_ON(ca->mi.state != BCH_MEMBER_STATE_RW);

	bch2_dev_allocator_add(c, ca);
	bch2_recalc_capacity(c);

	if (bch2_dev_allocator_start(ca))
		return "error starting allocator thread";

	if (bch2_copygc_start(c, ca))
		return "error starting copygc thread";

	return NULL;
}

int __bch2_dev_set_state(struct bch_fs *c, struct bch_dev *ca,
			 enum bch_member_state new_state, int flags)
{
	struct bch_sb_field_members *mi;
	int ret = 0;

	if (ca->mi.state == new_state)
		return 0;

	if (!bch2_dev_state_allowed(c, ca, new_state, flags))
		return -EINVAL;

	if (new_state != BCH_MEMBER_STATE_RW)
		__bch2_dev_read_only(c, ca);

	bch_notice(ca, "%s", bch2_dev_state[new_state]);

	mutex_lock(&c->sb_lock);
	mi = bch2_sb_get_members(c->disk_sb.sb);
	SET_BCH_MEMBER_STATE(&mi->members[ca->dev_idx], new_state);
	bch2_write_super(c);
	mutex_unlock(&c->sb_lock);

	if (new_state == BCH_MEMBER_STATE_RW &&
	    __bch2_dev_read_write(c, ca))
		ret = -ENOMEM;

	rebalance_wakeup(c);

	return ret;
}

int bch2_dev_set_state(struct bch_fs *c, struct bch_dev *ca,
		       enum bch_member_state new_state, int flags)
{
	int ret;

	mutex_lock(&c->state_lock);
	ret = __bch2_dev_set_state(c, ca, new_state, flags);
	mutex_unlock(&c->state_lock);

	return ret;
}

/* Device add/removal: */

int bch2_dev_remove(struct bch_fs *c, struct bch_dev *ca, int flags)
{
	struct bch_sb_field_members *mi;
	unsigned dev_idx = ca->dev_idx, data;
	int ret = -EINVAL;

	mutex_lock(&c->state_lock);

	percpu_ref_put(&ca->ref); /* XXX */

	if (!bch2_dev_state_allowed(c, ca, BCH_MEMBER_STATE_FAILED, flags)) {
		bch_err(ca, "Cannot remove without losing data");
		goto err;
	}

	__bch2_dev_read_only(c, ca);

	/*
	 * XXX: verify that dev_idx is really not in use anymore, anywhere
	 *
	 * flag_data_bad() does not check btree pointers
	 */
	ret = bch2_dev_data_drop(c, ca->dev_idx, flags);
	if (ret) {
		bch_err(ca, "Remove failed: error %i dropping data", ret);
		goto err;
	}

	ret = bch2_journal_flush_device_pins(&c->journal, ca->dev_idx);
	if (ret) {
		bch_err(ca, "Remove failed: error %i flushing journal", ret);
		goto err;
	}

	data = bch2_dev_has_data(c, ca);
	if (data) {
		char data_has_str[100];
1353

1354 1355
		bch2_flags_to_text(&PBUF(data_has_str),
				   bch2_data_types, data);
1356 1357 1358 1359 1360 1361 1362 1363
		bch_err(ca, "Remove failed, still has data (%s)", data_has_str);
		ret = -EBUSY;
		goto err;
	}

	ret = bch2_btree_delete_range(c, BTREE_ID_ALLOC,
				      POS(ca->dev_idx, 0),
				      POS(ca->dev_idx + 1, 0),
1364
				      NULL);
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
	if (ret) {
		bch_err(ca, "Remove failed, error deleting alloc info");
		goto err;
	}

	/*
	 * must flush all existing journal entries, they might have
	 * (overwritten) keys that point to the device we're removing:
	 */
	bch2_journal_flush_all_pins(&c->journal);
	ret = bch2_journal_error(&c->journal);
	if (ret) {
		bch_err(ca, "Remove failed, journal error");
		goto err;
	}

	__bch2_dev_offline(c, ca);

	mutex_lock(&c->sb_lock);
	rcu_assign_pointer(c->devs[ca->dev_idx], NULL);
	mutex_unlock(&c->sb_lock);

	percpu_ref_kill(&ca->ref);
	wait_for_completion(&ca->ref_completion);

	bch2_dev_free(ca);

	/*
	 * Free this device's slot in the bch_member array - all pointers to
	 * this device must be gone:
	 */
	mutex_lock(&c->sb_lock);
	mi = bch2_sb_get_members(c->disk_sb.sb);
	memset(&mi->members[dev_idx].uuid, 0, sizeof(mi->members[dev_idx].uuid));

	bch2_write_super(c);

	mutex_unlock(&c->sb_lock);
	mutex_unlock(&c->state_lock);
	return 0;
err:
1406 1407
	if (ca->mi.state == BCH_MEMBER_STATE_RW &&
	    !percpu_ref_is_zero(&ca->io_ref))
1408 1409 1410 1411 1412
		__bch2_dev_read_write(c, ca);
	mutex_unlock(&c->state_lock);
	return ret;
}

1413 1414 1415 1416 1417 1418 1419
static void dev_usage_clear(struct bch_dev *ca)
{
	struct bucket_array *buckets;
	int cpu;

	for_each_possible_cpu(cpu) {
		struct bch_dev_usage *p =
1420
			per_cpu_ptr(ca->usage[0], cpu);
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430
		memset(p, 0, sizeof(*p));
	}

	down_read(&ca->bucket_lock);
	buckets = bucket_array(ca);

	memset(buckets->b, 0, sizeof(buckets->b[0]) * buckets->nbuckets);
	up_read(&ca->bucket_lock);
}

1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
/* Add new device to running filesystem: */
int bch2_dev_add(struct bch_fs *c, const char *path)
{
	struct bch_opts opts = bch2_opts_empty();
	struct bch_sb_handle sb;
	const char *err;
	struct bch_dev *ca = NULL;
	struct bch_sb_field_members *mi;
	struct bch_member dev_mi;
	unsigned dev_idx, nr_devices, u64s;
	int ret;

	ret = bch2_read_super(path, &opts, &sb);
	if (ret)
		return ret;

	err = bch2_sb_validate(&sb);
	if (err)
		return -EINVAL;

	dev_mi = bch2_sb_get_members(sb.sb)->members[sb.sb->dev_idx];

	err = bch2_dev_may_add(sb.sb, c);
	if (err)
		return -EINVAL;

	ca = __bch2_dev_alloc(c, &dev_mi);
	if (!ca) {
		bch2_free_super(&sb);
		return -ENOMEM;
	}

	ret = __bch2_dev_attach_bdev(ca, &sb);
	if (ret) {
		bch2_dev_free(ca);
		return ret;
	}

1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480
	/*
	 * We want to allocate journal on the new device before adding the new
	 * device to the filesystem because allocating after we attach requires
	 * spinning up the allocator thread, and the allocator thread requires
	 * doing btree writes, which if the existing devices are RO isn't going
	 * to work
	 *
	 * So we have to mark where the superblocks are, but marking allocated
	 * data normally updates the filesystem usage too, so we have to mark,
	 * allocate the journal, reset all the marks, then remark after we
	 * attach...
	 */
1481
	bch2_mark_dev_superblock(ca->fs, ca, 0);
1482

1483 1484 1485 1486 1487
	err = "journal alloc failed";
	ret = bch2_dev_journal_alloc(ca);
	if (ret)
		goto err;

1488 1489
	dev_usage_clear(ca);

1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
	mutex_lock(&c->state_lock);
	mutex_lock(&c->sb_lock);

	err = "insufficient space in new superblock";
	ret = bch2_sb_from_fs(c, ca);
	if (ret)
		goto err_unlock;

	mi = bch2_sb_get_members(ca->disk_sb.sb);

	if (!bch2_sb_resize_members(&ca->disk_sb,
				le32_to_cpu(mi->field.u64s) +
				sizeof(dev_mi) / sizeof(u64))) {
		ret = -ENOSPC;
		goto err_unlock;
	}

	if (dynamic_fault("bcachefs:add:no_slot"))
		goto no_slot;

	mi = bch2_sb_get_members(c->disk_sb.sb);
	for (dev_idx = 0; dev_idx < BCH_SB_MEMBERS_MAX; dev_idx++)
		if (!bch2_dev_exists(c->disk_sb.sb, mi, dev_idx))
			goto have_slot;
no_slot:
	err = "no slots available in superblock";
	ret = -ENOSPC;
	goto err_unlock;

have_slot:
	nr_devices = max_t(unsigned, dev_idx + 1, c->sb.nr_devices);
	u64s = (sizeof(struct bch_sb_field_members) +
		sizeof(struct bch_member) * nr_devices) / sizeof(u64);

	err = "no space in superblock for member info";
	ret = -ENOSPC;

	mi = bch2_sb_resize_members(&c->disk_sb, u64s);
	if (!mi)
		goto err_unlock;

	/* success: */

	mi->members[dev_idx] = dev_mi;
1534
	mi->members[dev_idx].last_mount = cpu_to_le64(ktime_get_real_seconds());
1535 1536 1537 1538 1539
	c->disk_sb.sb->nr_devices	= nr_devices;

	ca->disk_sb.sb->dev_idx	= dev_idx;
	bch2_dev_attach(c, ca, dev_idx);

1540
	bch2_mark_dev_superblock(c, ca, 0);
1541

1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
	bch2_write_super(c);
	mutex_unlock(&c->sb_lock);

	if (ca->mi.state == BCH_MEMBER_STATE_RW) {
		err = __bch2_dev_read_write(c, ca);
		if (err)
			goto err_late;
	}

	mutex_unlock(&c->state_lock);
	return 0;

err_unlock:
	mutex_unlock(&c->sb_lock);
	mutex_unlock(&c->state_lock);
err:
	if (ca)
		bch2_dev_free(ca);
	bch2_free_super(&sb);
	bch_err(c, "Unable to add device: %s", err);
	return ret;
err_late:
	bch_err(c, "Error going rw after adding device: %s", err);
	return -EINVAL;
}

/* Hot add existing device to running filesystem: */
int bch2_dev_online(struct bch_fs *c, const char *path)
{
	struct bch_opts opts = bch2_opts_empty();
	struct bch_sb_handle sb = { NULL };
	struct bch_sb_field_members *mi;
	struct bch_dev *ca;
	unsigned dev_idx;
	const char *err;
	int ret;

	mutex_lock(&c->state_lock);

	ret = bch2_read_super(path, &opts, &sb);
	if (ret) {
		mutex_unlock(&c->state_lock);
		return ret;
	}

	dev_idx = sb.sb->dev_idx;

	err = bch2_dev_in_fs(c->disk_sb.sb, sb.sb);
	if (err)
		goto err;

	if (bch2_dev_attach_bdev(c, &sb)) {
		err = "bch2_dev_attach_bdev() error";
		goto err;
	}

	ca = bch_dev_locked(c, dev_idx);
	if (ca->mi.state == BCH_MEMBER_STATE_RW) {
		err = __bch2_dev_read_write(c, ca);
		if (err)
			goto err;
	}

	mutex_lock(&c->sb_lock);
	mi = bch2_sb_get_members(c->disk_sb.sb);

	mi->members[ca->dev_idx].last_mount =
1609
		cpu_to_le64(ktime_get_real_seconds());
1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 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 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 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 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904

	bch2_write_super(c);
	mutex_unlock(&c->sb_lock);

	mutex_unlock(&c->state_lock);
	return 0;
err:
	mutex_unlock(&c->state_lock);
	bch2_free_super(&sb);
	bch_err(c, "error bringing %s online: %s", path, err);
	return -EINVAL;
}

int bch2_dev_offline(struct bch_fs *c, struct bch_dev *ca, int flags)
{
	mutex_lock(&c->state_lock);

	if (!bch2_dev_is_online(ca)) {
		bch_err(ca, "Already offline");
		mutex_unlock(&c->state_lock);
		return 0;
	}

	if (!bch2_dev_state_allowed(c, ca, BCH_MEMBER_STATE_FAILED, flags)) {
		bch_err(ca, "Cannot offline required disk");
		mutex_unlock(&c->state_lock);
		return -EINVAL;
	}

	__bch2_dev_offline(c, ca);

	mutex_unlock(&c->state_lock);
	return 0;
}

int bch2_dev_resize(struct bch_fs *c, struct bch_dev *ca, u64 nbuckets)
{
	struct bch_member *mi;
	int ret = 0;

	mutex_lock(&c->state_lock);

	if (nbuckets < ca->mi.nbuckets) {
		bch_err(ca, "Cannot shrink yet");
		ret = -EINVAL;
		goto err;
	}

	if (bch2_dev_is_online(ca) &&
	    get_capacity(ca->disk_sb.bdev->bd_disk) <
	    ca->mi.bucket_size * nbuckets) {
		bch_err(ca, "New size larger than device");
		ret = -EINVAL;
		goto err;
	}

	ret = bch2_dev_buckets_resize(c, ca, nbuckets);
	if (ret) {
		bch_err(ca, "Resize error: %i", ret);
		goto err;
	}

	mutex_lock(&c->sb_lock);
	mi = &bch2_sb_get_members(c->disk_sb.sb)->members[ca->dev_idx];
	mi->nbuckets = cpu_to_le64(nbuckets);

	bch2_write_super(c);
	mutex_unlock(&c->sb_lock);

	bch2_recalc_capacity(c);
err:
	mutex_unlock(&c->state_lock);
	return ret;
}

/* return with ref on ca->ref: */
struct bch_dev *bch2_dev_lookup(struct bch_fs *c, const char *path)
{

	struct bch_dev *ca;
	dev_t dev;
	unsigned i;
	int ret;

	ret = lookup_bdev(path, &dev);
	if (ret)
		return ERR_PTR(ret);

	for_each_member_device(ca, c, i)
		if (ca->disk_sb.bdev->bd_dev == dev)
			goto found;

	ca = ERR_PTR(-ENOENT);
found:
	return ca;
}

/* Filesystem open: */

struct bch_fs *bch2_fs_open(char * const *devices, unsigned nr_devices,
			    struct bch_opts opts)
{
	struct bch_sb_handle *sb = NULL;
	struct bch_fs *c = NULL;
	unsigned i, best_sb = 0;
	const char *err;
	int ret = -ENOMEM;

	pr_verbose_init(opts, "");

	if (!nr_devices) {
		c = ERR_PTR(-EINVAL);
		goto out2;
	}

	if (!try_module_get(THIS_MODULE)) {
		c = ERR_PTR(-ENODEV);
		goto out2;
	}

	sb = kcalloc(nr_devices, sizeof(*sb), GFP_KERNEL);
	if (!sb)
		goto err;

	for (i = 0; i < nr_devices; i++) {
		ret = bch2_read_super(devices[i], &opts, &sb[i]);
		if (ret)
			goto err;

		err = bch2_sb_validate(&sb[i]);
		if (err)
			goto err_print;
	}

	for (i = 1; i < nr_devices; i++)
		if (le64_to_cpu(sb[i].sb->seq) >
		    le64_to_cpu(sb[best_sb].sb->seq))
			best_sb = i;

	for (i = 0; i < nr_devices; i++) {
		err = bch2_dev_in_fs(sb[best_sb].sb, sb[i].sb);
		if (err)
			goto err_print;
	}

	ret = -ENOMEM;
	c = bch2_fs_alloc(sb[best_sb].sb, opts);
	if (!c)
		goto err;

	err = "bch2_dev_online() error";
	mutex_lock(&c->state_lock);
	for (i = 0; i < nr_devices; i++)
		if (bch2_dev_attach_bdev(c, &sb[i])) {
			mutex_unlock(&c->state_lock);
			goto err_print;
		}
	mutex_unlock(&c->state_lock);

	err = "insufficient devices";
	if (!bch2_fs_may_start(c))
		goto err_print;

	if (!c->opts.nostart) {
		err = bch2_fs_start(c);
		if (err)
			goto err_print;
	}
out:
	kfree(sb);
	module_put(THIS_MODULE);
out2:
	pr_verbose_init(opts, "ret %i", PTR_ERR_OR_ZERO(c));
	return c;
err_print:
	pr_err("bch_fs_open err opening %s: %s",
	       devices[0], err);
	ret = -EINVAL;
err:
	if (c)
		bch2_fs_stop(c);
	for (i = 0; i < nr_devices; i++)
		bch2_free_super(&sb[i]);
	c = ERR_PTR(ret);
	goto out;
}

static const char *__bch2_fs_open_incremental(struct bch_sb_handle *sb,
					      struct bch_opts opts)
{
	const char *err;
	struct bch_fs *c;
	bool allocated_fs = false;

	err = bch2_sb_validate(sb);
	if (err)
		return err;

	mutex_lock(&bch_fs_list_lock);
	c = __bch2_uuid_to_fs(sb->sb->uuid);
	if (c) {
		closure_get(&c->cl);

		err = bch2_dev_in_fs(c->disk_sb.sb, sb->sb);
		if (err)
			goto err;
	} else {
		c = bch2_fs_alloc(sb->sb, opts);
		err = "cannot allocate memory";
		if (!c)
			goto err;

		allocated_fs = true;
	}

	err = "bch2_dev_online() error";

	mutex_lock(&c->sb_lock);
	if (bch2_dev_attach_bdev(c, sb)) {
		mutex_unlock(&c->sb_lock);
		goto err;
	}
	mutex_unlock(&c->sb_lock);

	if (!c->opts.nostart && bch2_fs_may_start(c)) {
		err = bch2_fs_start(c);
		if (err)
			goto err;
	}

	closure_put(&c->cl);
	mutex_unlock(&bch_fs_list_lock);

	return NULL;
err:
	mutex_unlock(&bch_fs_list_lock);

	if (allocated_fs)
		bch2_fs_stop(c);
	else if (c)
		closure_put(&c->cl);

	return err;
}

const char *bch2_fs_open_incremental(const char *path)
{
	struct bch_sb_handle sb;
	struct bch_opts opts = bch2_opts_empty();
	const char *err;

	if (bch2_read_super(path, &opts, &sb))
		return "error reading superblock";

	err = __bch2_fs_open_incremental(&sb, opts);
	bch2_free_super(&sb);

	return err;
}

/* Global interfaces/init */

static void bcachefs_exit(void)
{
	bch2_debug_exit();
	bch2_vfs_exit();
	bch2_chardev_exit();
	if (bcachefs_kset)
		kset_unregister(bcachefs_kset);
}

static int __init bcachefs_init(void)
{
	bch2_bkey_pack_test();
	bch2_inode_pack_test();

	if (!(bcachefs_kset = kset_create_and_add("bcachefs", NULL, fs_kobj)) ||
	    bch2_chardev_init() ||
	    bch2_vfs_init() ||
	    bch2_debug_init())
		goto err;

	return 0;
err:
	bcachefs_exit();
	return -ENOMEM;
}

#define BCH_DEBUG_PARAM(name, description)			\
	bool bch2_##name;					\
	module_param_named(name, bch2_##name, bool, 0644);	\
	MODULE_PARM_DESC(name, description);
BCH_DEBUG_PARAMS()
#undef BCH_DEBUG_PARAM

1905
unsigned bch2_metadata_version = bcachefs_metadata_version_current;
1906 1907 1908 1909
module_param_named(version, bch2_metadata_version, uint, 0400);

module_exit(bcachefs_exit);
module_init(bcachefs_init);