diff options
Diffstat (limited to 'drivers/md/bcache/super.c')
| -rw-r--r-- | drivers/md/bcache/super.c | 2043 |
1 files changed, 1441 insertions, 602 deletions
diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c index 8352fad765f6..c17d4517af22 100644 --- a/drivers/md/bcache/super.c +++ b/drivers/md/bcache/super.c @@ -1,3 +1,4 @@ +// SPDX-License-Identifier: GPL-2.0 /* * bcache setup/teardown code, and some metadata io - read a superblock and * figure out what to do with it. @@ -12,20 +13,21 @@ #include "extents.h" #include "request.h" #include "writeback.h" +#include "features.h" #include <linux/blkdev.h> -#include <linux/buffer_head.h> +#include <linux/pagemap.h> #include <linux/debugfs.h> -#include <linux/genhd.h> #include <linux/idr.h> #include <linux/kthread.h> +#include <linux/workqueue.h> #include <linux/module.h> #include <linux/random.h> #include <linux/reboot.h> #include <linux/sysfs.h> -MODULE_LICENSE("GPL"); -MODULE_AUTHOR("Kent Overstreet <kent.overstreet@gmail.com>"); +unsigned int bch_cutoff_writeback; +unsigned int bch_cutoff_writeback_sync; static const char bcache_magic[] = { 0xc6, 0x85, 0x73, 0xf6, 0x4e, 0x1a, 0x45, 0xca, @@ -37,43 +39,143 @@ static const char invalid_uuid[] = { 0xc8, 0x50, 0xfc, 0x5e, 0xcb, 0x16, 0xcd, 0x99 }; -/* Default is -1; we skip past it for struct cached_dev's cache mode */ -const char * const bch_cache_modes[] = { - "default", - "writethrough", - "writeback", - "writearound", - "none", - NULL -}; - static struct kobject *bcache_kobj; struct mutex bch_register_lock; +bool bcache_is_reboot; LIST_HEAD(bch_cache_sets); static LIST_HEAD(uncached_devices); static int bcache_major; -static DEFINE_IDA(bcache_minor); +static DEFINE_IDA(bcache_device_idx); static wait_queue_head_t unregister_wait; struct workqueue_struct *bcache_wq; +struct workqueue_struct *bch_flush_wq; +struct workqueue_struct *bch_journal_wq; + #define BTREE_MAX_PAGES (256 * 1024 / PAGE_SIZE) -#define BCACHE_MINORS 16 /* partition support */ +/* limitation of partitions number on single bcache device */ +#define BCACHE_MINORS 128 +/* limitation of bcache devices number on single system */ +#define BCACHE_DEVICE_IDX_MAX ((1U << MINORBITS)/BCACHE_MINORS) /* Superblock */ +static unsigned int get_bucket_size(struct cache_sb *sb, struct cache_sb_disk *s) +{ + unsigned int bucket_size = le16_to_cpu(s->bucket_size); + + if (sb->version >= BCACHE_SB_VERSION_CDEV_WITH_FEATURES) { + if (bch_has_feature_large_bucket(sb)) { + unsigned int max, order; + + max = sizeof(unsigned int) * BITS_PER_BYTE - 1; + order = le16_to_cpu(s->bucket_size); + /* + * bcache tool will make sure the overflow won't + * happen, an error message here is enough. + */ + if (order > max) + pr_err("Bucket size (1 << %u) overflows\n", + order); + bucket_size = 1 << order; + } else if (bch_has_feature_obso_large_bucket(sb)) { + bucket_size += + le16_to_cpu(s->obso_bucket_size_hi) << 16; + } + } + + return bucket_size; +} + +static const char *read_super_common(struct cache_sb *sb, struct block_device *bdev, + struct cache_sb_disk *s) +{ + const char *err; + unsigned int i; + + sb->first_bucket= le16_to_cpu(s->first_bucket); + sb->nbuckets = le64_to_cpu(s->nbuckets); + sb->bucket_size = get_bucket_size(sb, s); + + sb->nr_in_set = le16_to_cpu(s->nr_in_set); + sb->nr_this_dev = le16_to_cpu(s->nr_this_dev); + + err = "Too many journal buckets"; + if (sb->keys > SB_JOURNAL_BUCKETS) + goto err; + + err = "Too many buckets"; + if (sb->nbuckets > LONG_MAX) + goto err; + + err = "Not enough buckets"; + if (sb->nbuckets < 1 << 7) + goto err; + + err = "Bad block size (not power of 2)"; + if (!is_power_of_2(sb->block_size)) + goto err; + + err = "Bad block size (larger than page size)"; + if (sb->block_size > PAGE_SECTORS) + goto err; + + err = "Bad bucket size (not power of 2)"; + if (!is_power_of_2(sb->bucket_size)) + goto err; + + err = "Bad bucket size (smaller than page size)"; + if (sb->bucket_size < PAGE_SECTORS) + goto err; + + err = "Invalid superblock: device too small"; + if (get_capacity(bdev->bd_disk) < + sb->bucket_size * sb->nbuckets) + goto err; + + err = "Bad UUID"; + if (bch_is_zero(sb->set_uuid, 16)) + goto err; + + err = "Bad cache device number in set"; + if (!sb->nr_in_set || + sb->nr_in_set <= sb->nr_this_dev || + sb->nr_in_set > MAX_CACHES_PER_SET) + goto err; + + err = "Journal buckets not sequential"; + for (i = 0; i < sb->keys; i++) + if (sb->d[i] != sb->first_bucket + i) + goto err; + + err = "Too many journal buckets"; + if (sb->first_bucket + sb->keys > sb->nbuckets) + goto err; + + err = "Invalid superblock: first bucket comes before end of super"; + if (sb->first_bucket * sb->bucket_size < 16) + goto err; + + err = NULL; +err: + return err; +} + + static const char *read_super(struct cache_sb *sb, struct block_device *bdev, - struct page **res) + struct cache_sb_disk **res) { const char *err; - struct cache_sb *s; - struct buffer_head *bh = __bread(bdev, 1, SB_SIZE); - unsigned i; + struct cache_sb_disk *s; + struct folio *folio; + unsigned int i; - if (!bh) + folio = mapping_read_folio_gfp(bdev->bd_mapping, + SB_OFFSET >> PAGE_SHIFT, GFP_KERNEL); + if (IS_ERR(folio)) return "IO error"; - - s = (struct cache_sb *) bh->b_data; + s = folio_address(folio) + offset_in_folio(folio, SB_OFFSET); sb->offset = le64_to_cpu(s->offset); sb->version = le64_to_cpu(s->version); @@ -86,26 +188,22 @@ static const char *read_super(struct cache_sb *sb, struct block_device *bdev, sb->flags = le64_to_cpu(s->flags); sb->seq = le64_to_cpu(s->seq); sb->last_mount = le32_to_cpu(s->last_mount); - sb->first_bucket = le16_to_cpu(s->first_bucket); sb->keys = le16_to_cpu(s->keys); for (i = 0; i < SB_JOURNAL_BUCKETS; i++) sb->d[i] = le64_to_cpu(s->d[i]); - pr_debug("read sb version %llu, flags %llu, seq %llu, journal size %u", + pr_debug("read sb version %llu, flags %llu, seq %llu, journal size %u\n", sb->version, sb->flags, sb->seq, sb->keys); - err = "Not a bcache superblock"; + err = "Not a bcache superblock (bad offset)"; if (sb->offset != SB_SECTOR) goto err; + err = "Not a bcache superblock (bad magic)"; if (memcmp(sb->magic, bcache_magic, 16)) goto err; - err = "Too many journal buckets"; - if (sb->keys > SB_JOURNAL_BUCKETS) - goto err; - err = "Bad checksum"; if (s->csum != csum_set(s)) goto err; @@ -125,6 +223,7 @@ static const char *read_super(struct cache_sb *sb, struct block_device *bdev, sb->data_offset = BDEV_DATA_START_DEFAULT; break; case BCACHE_SB_VERSION_BDEV_WITH_OFFSET: + case BCACHE_SB_VERSION_BDEV_WITH_FEATURES: sb->data_offset = le64_to_cpu(s->data_offset); err = "Bad data offset"; @@ -134,90 +233,69 @@ static const char *read_super(struct cache_sb *sb, struct block_device *bdev, break; case BCACHE_SB_VERSION_CDEV: case BCACHE_SB_VERSION_CDEV_WITH_UUID: - sb->nbuckets = le64_to_cpu(s->nbuckets); - sb->bucket_size = le16_to_cpu(s->bucket_size); - - sb->nr_in_set = le16_to_cpu(s->nr_in_set); - sb->nr_this_dev = le16_to_cpu(s->nr_this_dev); - - err = "Too many buckets"; - if (sb->nbuckets > LONG_MAX) - goto err; - - err = "Not enough buckets"; - if (sb->nbuckets < 1 << 7) - goto err; - - err = "Bad block/bucket size"; - if (!is_power_of_2(sb->block_size) || - sb->block_size > PAGE_SECTORS || - !is_power_of_2(sb->bucket_size) || - sb->bucket_size < PAGE_SECTORS) - goto err; - - err = "Invalid superblock: device too small"; - if (get_capacity(bdev->bd_disk) < sb->bucket_size * sb->nbuckets) + err = read_super_common(sb, bdev, s); + if (err) goto err; + break; + case BCACHE_SB_VERSION_CDEV_WITH_FEATURES: + /* + * Feature bits are needed in read_super_common(), + * convert them firstly. + */ + sb->feature_compat = le64_to_cpu(s->feature_compat); + sb->feature_incompat = le64_to_cpu(s->feature_incompat); + sb->feature_ro_compat = le64_to_cpu(s->feature_ro_compat); - err = "Bad UUID"; - if (bch_is_zero(sb->set_uuid, 16)) + /* Check incompatible features */ + err = "Unsupported compatible feature found"; + if (bch_has_unknown_compat_features(sb)) goto err; - err = "Bad cache device number in set"; - if (!sb->nr_in_set || - sb->nr_in_set <= sb->nr_this_dev || - sb->nr_in_set > MAX_CACHES_PER_SET) + err = "Unsupported read-only compatible feature found"; + if (bch_has_unknown_ro_compat_features(sb)) goto err; - err = "Journal buckets not sequential"; - for (i = 0; i < sb->keys; i++) - if (sb->d[i] != sb->first_bucket + i) - goto err; - - err = "Too many journal buckets"; - if (sb->first_bucket + sb->keys > sb->nbuckets) + err = "Unsupported incompatible feature found"; + if (bch_has_unknown_incompat_features(sb)) goto err; - err = "Invalid superblock: first bucket comes before end of super"; - if (sb->first_bucket * sb->bucket_size < 16) + err = read_super_common(sb, bdev, s); + if (err) goto err; - break; default: err = "Unsupported superblock version"; goto err; } - sb->last_mount = get_seconds(); - err = NULL; - - get_page(bh->b_page); - *res = bh->b_page; + sb->last_mount = (u32)ktime_get_real_seconds(); + *res = s; + return NULL; err: - put_bh(bh); + folio_put(folio); return err; } static void write_bdev_super_endio(struct bio *bio) { struct cached_dev *dc = bio->bi_private; - /* XXX: error checking */ + + if (bio->bi_status) + bch_count_backing_io_errors(dc, bio); closure_put(&dc->sb_write); } -static void __write_super(struct cache_sb *sb, struct bio *bio) +static void __write_super(struct cache_sb *sb, struct cache_sb_disk *out, + struct bio *bio) { - struct cache_sb *out = page_address(bio->bi_io_vec[0].bv_page); - unsigned i; + unsigned int i; + bio->bi_opf = REQ_OP_WRITE | REQ_SYNC | REQ_META; bio->bi_iter.bi_sector = SB_SECTOR; - bio->bi_iter.bi_size = SB_SIZE; - bio_set_op_attrs(bio, REQ_OP_WRITE, REQ_SYNC|REQ_META); - bch_bio_map(bio, NULL); + bio_add_virt_nofail(bio, out, SB_SIZE); out->offset = cpu_to_le64(sb->offset); - out->version = cpu_to_le64(sb->version); memcpy(out->uuid, sb->uuid, 16); memcpy(out->set_uuid, sb->set_uuid, 16); @@ -233,17 +311,24 @@ static void __write_super(struct cache_sb *sb, struct bio *bio) for (i = 0; i < sb->keys; i++) out->d[i] = cpu_to_le64(sb->d[i]); + if (sb->version >= BCACHE_SB_VERSION_CDEV_WITH_FEATURES) { + out->feature_compat = cpu_to_le64(sb->feature_compat); + out->feature_incompat = cpu_to_le64(sb->feature_incompat); + out->feature_ro_compat = cpu_to_le64(sb->feature_ro_compat); + } + + out->version = cpu_to_le64(sb->version); out->csum = csum_set(out); - pr_debug("ver %llu, flags %llu, seq %llu", + pr_debug("ver %llu, flags %llu, seq %llu\n", sb->version, sb->flags, sb->seq); submit_bio(bio); } -static void bch_write_bdev_super_unlock(struct closure *cl) +static CLOSURE_CALLBACK(bch_write_bdev_super_unlock) { - struct cached_dev *dc = container_of(cl, struct cached_dev, sb_write); + closure_type(dc, struct cached_dev, sb_write); up(&dc->sb_write_mutex); } @@ -256,13 +341,13 @@ void bch_write_bdev_super(struct cached_dev *dc, struct closure *parent) down(&dc->sb_write_mutex); closure_init(cl, parent); - bio_reset(bio); - bio->bi_bdev = dc->bdev; + bio_init(bio, dc->bdev, dc->sb_bv, 1, 0); bio->bi_end_io = write_bdev_super_endio; bio->bi_private = dc; closure_get(cl); - __write_super(&dc->sb, bio); + /* I/O request sent to backing device */ + __write_super(&dc->sb, dc->sb_disk, bio); closure_return_with_destructor(cl, bch_write_bdev_super_unlock); } @@ -271,13 +356,15 @@ static void write_super_endio(struct bio *bio) { struct cache *ca = bio->bi_private; - bch_count_io_errors(ca, bio->bi_status, "writing superblock"); + /* is_read = 0 */ + bch_count_io_errors(ca, bio->bi_status, 0, + "writing superblock"); closure_put(&ca->set->sb_write); } -static void bcache_write_super_unlock(struct closure *cl) +static CLOSURE_CALLBACK(bcache_write_super_unlock) { - struct cache_set *c = container_of(cl, struct cache_set, sb_write); + closure_type(c, struct cache_set, sb_write); up(&c->sb_write_mutex); } @@ -285,31 +372,24 @@ static void bcache_write_super_unlock(struct closure *cl) void bcache_write_super(struct cache_set *c) { struct closure *cl = &c->sb_write; - struct cache *ca; - unsigned i; + struct cache *ca = c->cache; + struct bio *bio = &ca->sb_bio; + unsigned int version = BCACHE_SB_VERSION_CDEV_WITH_UUID; down(&c->sb_write_mutex); closure_init(cl, &c->cl); - c->sb.seq++; + ca->sb.seq++; - for_each_cache(ca, c, i) { - struct bio *bio = &ca->sb_bio; + if (ca->sb.version < version) + ca->sb.version = version; - ca->sb.version = BCACHE_SB_VERSION_CDEV_WITH_UUID; - ca->sb.seq = c->sb.seq; - ca->sb.last_mount = c->sb.last_mount; - - SET_CACHE_SYNC(&ca->sb, CACHE_SYNC(&c->sb)); - - bio_reset(bio); - bio->bi_bdev = ca->bdev; - bio->bi_end_io = write_super_endio; - bio->bi_private = ca; + bio_init(bio, ca->bdev, ca->sb_bv, 1, 0); + bio->bi_end_io = write_super_endio; + bio->bi_private = ca; - closure_get(cl); - __write_super(&ca->sb, bio); - } + closure_get(cl); + __write_super(&ca->sb, ca->sb_disk, bio); closure_return_with_destructor(cl, bcache_write_super_unlock); } @@ -326,19 +406,19 @@ static void uuid_endio(struct bio *bio) closure_put(cl); } -static void uuid_io_unlock(struct closure *cl) +static CLOSURE_CALLBACK(uuid_io_unlock) { - struct cache_set *c = container_of(cl, struct cache_set, uuid_write); + closure_type(c, struct cache_set, uuid_write); up(&c->uuid_write_mutex); } -static void uuid_io(struct cache_set *c, int op, unsigned long op_flags, - struct bkey *k, struct closure *parent) +static void uuid_io(struct cache_set *c, blk_opf_t opf, struct bkey *k, + struct closure *parent) { struct closure *cl = &c->uuid_write; struct uuid_entry *u; - unsigned i; + unsigned int i; char buf[80]; BUG_ON(!parent); @@ -348,26 +428,26 @@ static void uuid_io(struct cache_set *c, int op, unsigned long op_flags, for (i = 0; i < KEY_PTRS(k); i++) { struct bio *bio = bch_bbio_alloc(c); - bio->bi_opf = REQ_SYNC | REQ_META | op_flags; + bio->bi_opf = opf | REQ_SYNC | REQ_META; bio->bi_iter.bi_size = KEY_SIZE(k) << 9; bio->bi_end_io = uuid_endio; bio->bi_private = cl; - bio_set_op_attrs(bio, op, REQ_SYNC|REQ_META|op_flags); bch_bio_map(bio, c->uuids); bch_submit_bbio(bio, c, k, i); - if (op != REQ_OP_WRITE) + if ((opf & REQ_OP_MASK) != REQ_OP_WRITE) break; } bch_extent_to_text(buf, sizeof(buf), k); - pr_debug("%s UUIDs at %s", op == REQ_OP_WRITE ? "wrote" : "read", buf); + pr_debug("%s UUIDs at %s\n", (opf & REQ_OP_MASK) == REQ_OP_WRITE ? + "wrote" : "read", buf); for (u = c->uuids; u < c->uuids + c->nr_uuids; u++) if (!bch_is_zero(u->uuid, 16)) - pr_debug("Slot %zi: %pU: %s: 1st: %u last: %u inv: %u", + pr_debug("Slot %zi: %pU: %s: 1st: %u last: %u inv: %u\n", u - c->uuids, u->uuid, u->label, u->first_reg, u->last_reg, u->invalidated); @@ -382,7 +462,7 @@ static char *uuid_read(struct cache_set *c, struct jset *j, struct closure *cl) return "bad uuid pointer"; bkey_copy(&c->uuid_bucket, k); - uuid_io(c, REQ_OP_READ, 0, k, cl); + uuid_io(c, REQ_OP_READ, k, cl); if (j->version < BCACHE_JSET_VERSION_UUIDv1) { struct uuid_entry_v0 *u0 = (void *) c->uuids; @@ -419,17 +499,23 @@ static int __uuid_write(struct cache_set *c) { BKEY_PADDED(key) k; struct closure cl; - closure_init_stack(&cl); + struct cache *ca = c->cache; + unsigned int size; + closure_init_stack(&cl); lockdep_assert_held(&bch_register_lock); - if (bch_bucket_alloc_set(c, RESERVE_BTREE, &k.key, 1, true)) + if (bch_bucket_alloc_set(c, RESERVE_BTREE, &k.key, true)) return 1; - SET_KEY_SIZE(&k.key, c->sb.bucket_size); - uuid_io(c, REQ_OP_WRITE, 0, &k.key, &cl); + size = meta_bucket_pages(&ca->sb) * PAGE_SECTORS; + SET_KEY_SIZE(&k.key, size); + uuid_io(c, REQ_OP_WRITE, &k.key, &cl); closure_sync(&cl); + /* Only one bucket used for uuid write */ + atomic_long_add(ca->sb.bucket_size, &ca->meta_sectors_written); + bkey_copy(&c->uuid_bucket, &k.key); bkey_put(c, &k.key); return 0; @@ -459,7 +545,9 @@ static struct uuid_entry *uuid_find(struct cache_set *c, const char *uuid) static struct uuid_entry *uuid_find_empty(struct cache_set *c) { - static const char zero_uuid[16] = "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"; + static const char zero_uuid[16] __nonstring = + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + return uuid_find(c, zero_uuid); } @@ -467,8 +555,8 @@ static struct uuid_entry *uuid_find_empty(struct cache_set *c) * Bucket priorities/gens: * * For each bucket, we store on disk its - * 8 bit gen - * 16 bit priority + * 8 bit gen + * 16 bit priority * * See alloc.c for an explanation of the gen. The priority is used to implement * lru (and in the future other) cache replacement policies; for most purposes @@ -499,8 +587,7 @@ static void prio_endio(struct bio *bio) closure_put(&ca->prio); } -static void prio_io(struct cache *ca, uint64_t bucket, int op, - unsigned long op_flags) +static void prio_io(struct cache *ca, uint64_t bucket, blk_opf_t opf) { struct closure *cl = &ca->prio; struct bio *bio = bch_bbio_alloc(ca->set); @@ -508,24 +595,41 @@ static void prio_io(struct cache *ca, uint64_t bucket, int op, closure_init_stack(cl); bio->bi_iter.bi_sector = bucket * ca->sb.bucket_size; - bio->bi_bdev = ca->bdev; - bio->bi_iter.bi_size = bucket_bytes(ca); + bio_set_dev(bio, ca->bdev); + bio->bi_iter.bi_size = meta_bucket_bytes(&ca->sb); bio->bi_end_io = prio_endio; bio->bi_private = ca; - bio_set_op_attrs(bio, op, REQ_SYNC|REQ_META|op_flags); + bio->bi_opf = opf | REQ_SYNC | REQ_META; bch_bio_map(bio, ca->disk_buckets); - closure_bio_submit(bio, &ca->prio); + closure_bio_submit(ca->set, bio, &ca->prio); closure_sync(cl); } -void bch_prio_write(struct cache *ca) +int bch_prio_write(struct cache *ca, bool wait) { int i; struct bucket *b; struct closure cl; + pr_debug("free_prio=%zu, free_none=%zu, free_inc=%zu\n", + fifo_used(&ca->free[RESERVE_PRIO]), + fifo_used(&ca->free[RESERVE_NONE]), + fifo_used(&ca->free_inc)); + + /* + * Pre-check if there are enough free buckets. In the non-blocking + * scenario it's better to fail early rather than starting to allocate + * buckets and do a cleanup later in case of failure. + */ + if (!wait) { + size_t avail = fifo_used(&ca->free[RESERVE_PRIO]) + + fifo_used(&ca->free[RESERVE_NONE]); + if (prio_buckets(ca) > avail) + return -ENOMEM; + } + closure_init_stack(&cl); lockdep_assert_held(&ca->set->bucket_lock); @@ -535,9 +639,6 @@ void bch_prio_write(struct cache *ca) atomic_long_add(ca->sb.bucket_size * prio_buckets(ca), &ca->meta_sectors_written); - //pr_debug("free %zu, free_inc %zu, unused %zu", fifo_used(&ca->free), - // fifo_used(&ca->free_inc), fifo_used(&ca->unused)); - for (i = prio_buckets(ca) - 1; i >= 0; --i) { long bucket; struct prio_set *p = ca->disk_buckets; @@ -553,13 +654,13 @@ void bch_prio_write(struct cache *ca) p->next_bucket = ca->prio_buckets[i + 1]; p->magic = pset_magic(&ca->sb); - p->csum = bch_crc64(&p->magic, bucket_bytes(ca) - 8); + p->csum = bch_crc64(&p->magic, meta_bucket_bytes(&ca->sb) - 8); - bucket = bch_bucket_alloc(ca, RESERVE_PRIO, true); + bucket = bch_bucket_alloc(ca, RESERVE_PRIO, wait); BUG_ON(bucket == -1); mutex_unlock(&ca->set->bucket_lock); - prio_io(ca, bucket, REQ_OP_WRITE, 0); + prio_io(ca, bucket, REQ_OP_WRITE); mutex_lock(&ca->set->bucket_lock); ca->prio_buckets[i] = bucket; @@ -584,14 +685,16 @@ void bch_prio_write(struct cache *ca) ca->prio_last_buckets[i] = ca->prio_buckets[i]; } + return 0; } -static void prio_read(struct cache *ca, uint64_t bucket) +static int prio_read(struct cache *ca, uint64_t bucket) { struct prio_set *p = ca->disk_buckets; struct bucket_disk *d = p->data + prios_per_bucket(ca), *end = d; struct bucket *b; - unsigned bucket_nr = 0; + unsigned int bucket_nr = 0; + int ret = -EIO; for (b = ca->buckets; b < ca->buckets + ca->sb.nbuckets; @@ -601,13 +704,18 @@ static void prio_read(struct cache *ca, uint64_t bucket) ca->prio_last_buckets[bucket_nr] = bucket; bucket_nr++; - prio_io(ca, bucket, REQ_OP_READ, 0); + prio_io(ca, bucket, REQ_OP_READ); - if (p->csum != bch_crc64(&p->magic, bucket_bytes(ca) - 8)) - pr_warn("bad csum reading priorities"); + if (p->csum != + bch_crc64(&p->magic, meta_bucket_bytes(&ca->sb) - 8)) { + pr_warn("bad csum reading priorities\n"); + goto out; + } - if (p->magic != pset_magic(&ca->sb)) - pr_warn("bad magic reading priorities"); + if (p->magic != pset_magic(&ca->sb)) { + pr_warn("bad magic reading priorities\n"); + goto out; + } bucket = p->next_bucket; d = p->data; @@ -616,13 +724,18 @@ static void prio_read(struct cache *ca, uint64_t bucket) b->prio = le16_to_cpu(d->prio); b->gen = b->last_gc = d->gen; } + + ret = 0; +out: + return ret; } /* Bcache device */ -static int open_dev(struct block_device *b, fmode_t mode) +static int open_dev(struct gendisk *disk, blk_mode_t mode) { - struct bcache_device *d = b->bd_disk->private_data; + struct bcache_device *d = disk->private_data; + if (test_bit(BCACHE_DEV_CLOSING, &d->flags)) return -ENXIO; @@ -630,20 +743,31 @@ static int open_dev(struct block_device *b, fmode_t mode) return 0; } -static void release_dev(struct gendisk *b, fmode_t mode) +static void release_dev(struct gendisk *b) { struct bcache_device *d = b->private_data; + closure_put(&d->cl); } -static int ioctl_dev(struct block_device *b, fmode_t mode, +static int ioctl_dev(struct block_device *b, blk_mode_t mode, unsigned int cmd, unsigned long arg) { struct bcache_device *d = b->bd_disk->private_data; + return d->ioctl(d, mode, cmd, arg); } -static const struct block_device_operations bcache_ops = { +static const struct block_device_operations bcache_cached_ops = { + .submit_bio = cached_dev_submit_bio, + .open = open_dev, + .release = release_dev, + .ioctl = ioctl_dev, + .owner = THIS_MODULE, +}; + +static const struct block_device_operations bcache_flash_ops = { + .submit_bio = flash_dev_submit_bio, .open = open_dev, .release = release_dev, .ioctl = ioctl_dev, @@ -653,6 +777,11 @@ static const struct block_device_operations bcache_ops = { void bcache_device_stop(struct bcache_device *d) { if (!test_and_set_bit(BCACHE_DEV_CLOSING, &d->flags)) + /* + * closure_fn set to + * - cached device: cached_dev_flush() + * - flash dev: flash_dev_flush() + */ closure_queue(&d->cl); } @@ -661,32 +790,33 @@ static void bcache_device_unlink(struct bcache_device *d) lockdep_assert_held(&bch_register_lock); if (d->c && !test_and_set_bit(BCACHE_DEV_UNLINK_DONE, &d->flags)) { - unsigned i; - struct cache *ca; + struct cache *ca = d->c->cache; sysfs_remove_link(&d->c->kobj, d->name); sysfs_remove_link(&d->kobj, "cache"); - for_each_cache(ca, d->c, i) - bd_unlink_disk_holder(ca->bdev, d->disk); + bd_unlink_disk_holder(ca->bdev, d->disk); } } static void bcache_device_link(struct bcache_device *d, struct cache_set *c, const char *name) { - unsigned i; - struct cache *ca; + struct cache *ca = c->cache; + int ret; - for_each_cache(ca, d->c, i) - bd_link_disk_holder(ca->bdev, d->disk); + bd_link_disk_holder(ca->bdev, d->disk); snprintf(d->name, BCACHEDEVNAME_SIZE, "%s%u", name, d->id); - WARN(sysfs_create_link(&d->kobj, &c->kobj, "cache") || - sysfs_create_link(&c->kobj, &d->kobj, d->name), - "Couldn't create device <-> cache set symlinks"); + ret = sysfs_create_link(&d->kobj, &c->kobj, "cache"); + if (ret < 0) + pr_err("Couldn't create device -> cache set symlink\n"); + + ret = sysfs_create_link(&c->kobj, &d->kobj, d->name); + if (ret < 0) + pr_err("Couldn't create cache set -> device symlink\n"); clear_bit(BCACHE_DEV_UNLINK_DONE, &d->flags); } @@ -695,12 +825,14 @@ static void bcache_device_detach(struct bcache_device *d) { lockdep_assert_held(&bch_register_lock); + atomic_dec(&d->c->attached_dev_nr); + if (test_bit(BCACHE_DEV_DETACHING, &d->flags)) { struct uuid_entry *u = d->c->uuids + d->id; SET_UUID_FLASH_ONLY(u, 0); memcpy(u->uuid, invalid_uuid, 16); - u->invalidated = cpu_to_le32(get_seconds()); + u->invalidated = cpu_to_le32((u32)ktime_get_real_seconds()); bch_uuid_write(d->c); } @@ -712,59 +844,91 @@ static void bcache_device_detach(struct bcache_device *d) } static void bcache_device_attach(struct bcache_device *d, struct cache_set *c, - unsigned id) + unsigned int id) { d->id = id; d->c = c; c->devices[id] = d; + if (id >= c->devices_max_used) + c->devices_max_used = id + 1; + closure_get(&c->caching); } +static inline int first_minor_to_idx(int first_minor) +{ + return (first_minor/BCACHE_MINORS); +} + +static inline int idx_to_first_minor(int idx) +{ + return (idx * BCACHE_MINORS); +} + static void bcache_device_free(struct bcache_device *d) { + struct gendisk *disk = d->disk; + lockdep_assert_held(&bch_register_lock); - pr_info("%s stopped", d->disk->disk_name); + if (disk) + pr_info("%s stopped\n", disk->disk_name); + else + pr_err("bcache device (NULL gendisk) stopped\n"); if (d->c) bcache_device_detach(d); - if (d->disk && d->disk->flags & GENHD_FL_UP) - del_gendisk(d->disk); - if (d->disk && d->disk->queue) - blk_cleanup_queue(d->disk->queue); - if (d->disk) { - ida_simple_remove(&bcache_minor, d->disk->first_minor); - put_disk(d->disk); + + if (disk) { + ida_free(&bcache_device_idx, + first_minor_to_idx(disk->first_minor)); + put_disk(disk); } - if (d->bio_split) - bioset_free(d->bio_split); + bioset_exit(&d->bio_split); kvfree(d->full_dirty_stripes); kvfree(d->stripe_sectors_dirty); closure_debug_destroy(&d->cl); } -static int bcache_device_init(struct bcache_device *d, unsigned block_size, - sector_t sectors) -{ - struct request_queue *q; - size_t n; - int minor; +static int bcache_device_init(struct bcache_device *d, unsigned int block_size, + sector_t sectors, struct block_device *cached_bdev, + const struct block_device_operations *ops) +{ + const size_t max_stripes = min_t(size_t, INT_MAX, + SIZE_MAX / sizeof(atomic_t)); + struct queue_limits lim = { + .max_hw_sectors = UINT_MAX, + .max_sectors = UINT_MAX, + .max_segment_size = UINT_MAX, + .max_segments = BIO_MAX_VECS, + .max_hw_discard_sectors = UINT_MAX, + .io_min = block_size, + .logical_block_size = block_size, + .physical_block_size = block_size, + .features = BLK_FEAT_WRITE_CACHE | BLK_FEAT_FUA, + }; + uint64_t n; + int idx; + if (cached_bdev) { + d->stripe_size = bdev_io_opt(cached_bdev) >> SECTOR_SHIFT; + lim.io_opt = umax(block_size, bdev_io_opt(cached_bdev)); + } if (!d->stripe_size) d->stripe_size = 1 << 31; + else if (d->stripe_size < BCH_MIN_STRIPE_SZ) + d->stripe_size = roundup(BCH_MIN_STRIPE_SZ, d->stripe_size); - d->nr_stripes = DIV_ROUND_UP_ULL(sectors, d->stripe_size); - - if (!d->nr_stripes || - d->nr_stripes > INT_MAX || - d->nr_stripes > SIZE_MAX / sizeof(atomic_t)) { - pr_err("nr_stripes too large or invalid: %u (start sector beyond end of disk?)", - (unsigned)d->nr_stripes); + n = DIV_ROUND_UP_ULL(sectors, d->stripe_size); + if (!n || n > max_stripes) { + pr_err("nr_stripes too large or invalid: %llu (start sector beyond end of disk?)\n", + n); return -ENOMEM; } + d->nr_stripes = n; n = d->nr_stripes * sizeof(atomic_t); d->stripe_sectors_dirty = kvzalloc(n, GFP_KERNEL); @@ -774,54 +938,54 @@ static int bcache_device_init(struct bcache_device *d, unsigned block_size, n = BITS_TO_LONGS(d->nr_stripes) * sizeof(unsigned long); d->full_dirty_stripes = kvzalloc(n, GFP_KERNEL); if (!d->full_dirty_stripes) - return -ENOMEM; + goto out_free_stripe_sectors_dirty; - minor = ida_simple_get(&bcache_minor, 0, MINORMASK + 1, GFP_KERNEL); - if (minor < 0) - return minor; + idx = ida_alloc_max(&bcache_device_idx, BCACHE_DEVICE_IDX_MAX - 1, + GFP_KERNEL); + if (idx < 0) + goto out_free_full_dirty_stripes; - minor *= BCACHE_MINORS; + if (bioset_init(&d->bio_split, 4, offsetof(struct bbio, bio), + BIOSET_NEED_BVECS|BIOSET_NEED_RESCUER)) + goto out_ida_remove; - if (!(d->bio_split = bioset_create(4, offsetof(struct bbio, bio), - BIOSET_NEED_BVECS | - BIOSET_NEED_RESCUER)) || - !(d->disk = alloc_disk(BCACHE_MINORS))) { - ida_simple_remove(&bcache_minor, minor); - return -ENOMEM; + if (lim.logical_block_size > PAGE_SIZE && cached_bdev) { + /* + * This should only happen with BCACHE_SB_VERSION_BDEV. + * Block/page size is checked for BCACHE_SB_VERSION_CDEV. + */ + pr_info("bcache%i: sb/logical block size (%u) greater than page size (%lu) falling back to device logical block size (%u)\n", + idx, lim.logical_block_size, + PAGE_SIZE, bdev_logical_block_size(cached_bdev)); + + /* This also adjusts physical block size/min io size if needed */ + lim.logical_block_size = bdev_logical_block_size(cached_bdev); } + d->disk = blk_alloc_disk(&lim, NUMA_NO_NODE); + if (IS_ERR(d->disk)) + goto out_bioset_exit; + set_capacity(d->disk, sectors); - snprintf(d->disk->disk_name, DISK_NAME_LEN, "bcache%i", minor); + snprintf(d->disk->disk_name, DISK_NAME_LEN, "bcache%i", idx); d->disk->major = bcache_major; - d->disk->first_minor = minor; - d->disk->fops = &bcache_ops; + d->disk->first_minor = idx_to_first_minor(idx); + d->disk->minors = BCACHE_MINORS; + d->disk->fops = ops; d->disk->private_data = d; + return 0; - q = blk_alloc_queue(GFP_KERNEL); - if (!q) - return -ENOMEM; - - blk_queue_make_request(q, NULL); - d->disk->queue = q; - q->queuedata = d; - q->backing_dev_info->congested_data = d; - q->limits.max_hw_sectors = UINT_MAX; - q->limits.max_sectors = UINT_MAX; - q->limits.max_segment_size = UINT_MAX; - q->limits.max_segments = BIO_MAX_PAGES; - blk_queue_max_discard_sectors(q, UINT_MAX); - q->limits.discard_granularity = 512; - q->limits.io_min = block_size; - q->limits.logical_block_size = block_size; - q->limits.physical_block_size = block_size; - set_bit(QUEUE_FLAG_NONROT, &d->disk->queue->queue_flags); - clear_bit(QUEUE_FLAG_ADD_RANDOM, &d->disk->queue->queue_flags); - set_bit(QUEUE_FLAG_DISCARD, &d->disk->queue->queue_flags); - - blk_queue_write_cache(q, true, true); +out_bioset_exit: + bioset_exit(&d->bio_split); +out_ida_remove: + ida_free(&bcache_device_idx, idx); +out_free_full_dirty_stripes: + kvfree(d->full_dirty_stripes); +out_free_stripe_sectors_dirty: + kvfree(d->stripe_sectors_dirty); + return -ENOMEM; - return 0; } /* Cached device */ @@ -832,35 +996,77 @@ static void calc_cached_dev_sectors(struct cache_set *c) struct cached_dev *dc; list_for_each_entry(dc, &c->cached_devs, list) - sectors += bdev_sectors(dc->bdev); + sectors += bdev_nr_sectors(dc->bdev); c->cached_dev_sectors = sectors; } -void bch_cached_dev_run(struct cached_dev *dc) +#define BACKING_DEV_OFFLINE_TIMEOUT 5 +static int cached_dev_status_update(void *arg) +{ + struct cached_dev *dc = arg; + struct request_queue *q; + + /* + * If this delayed worker is stopping outside, directly quit here. + * dc->io_disable might be set via sysfs interface, so check it + * here too. + */ + while (!kthread_should_stop() && !dc->io_disable) { + q = bdev_get_queue(dc->bdev); + if (blk_queue_dying(q)) + dc->offline_seconds++; + else + dc->offline_seconds = 0; + + if (dc->offline_seconds >= BACKING_DEV_OFFLINE_TIMEOUT) { + pr_err("%pg: device offline for %d seconds\n", + dc->bdev, + BACKING_DEV_OFFLINE_TIMEOUT); + pr_err("%s: disable I/O request due to backing device offline\n", + dc->disk.name); + dc->io_disable = true; + /* let others know earlier that io_disable is true */ + smp_mb(); + bcache_device_stop(&dc->disk); + break; + } + schedule_timeout_interruptible(HZ); + } + + wait_for_kthread_stop(); + return 0; +} + + +int bch_cached_dev_run(struct cached_dev *dc) { + int ret = 0; struct bcache_device *d = &dc->disk; - char buf[SB_LABEL_SIZE + 1]; + char *buf = kmemdup_nul(dc->sb.label, SB_LABEL_SIZE, GFP_KERNEL); char *env[] = { "DRIVER=bcache", kasprintf(GFP_KERNEL, "CACHED_UUID=%pU", dc->sb.uuid), - NULL, + kasprintf(GFP_KERNEL, "CACHED_LABEL=%s", buf ? : ""), NULL, }; - memcpy(buf, dc->sb.label, SB_LABEL_SIZE); - buf[SB_LABEL_SIZE] = '\0'; - env[2] = kasprintf(GFP_KERNEL, "CACHED_LABEL=%s", buf); + if (dc->io_disable) { + pr_err("I/O disabled on cached dev %pg\n", dc->bdev); + ret = -EIO; + goto out; + } if (atomic_xchg(&dc->running, 1)) { - kfree(env[1]); - kfree(env[2]); - return; + pr_info("cached dev %pg is running already\n", dc->bdev); + ret = -EBUSY; + goto out; } if (!d->c && BDEV_STATE(&dc->sb) != BDEV_STATE_NONE) { struct closure cl; + closure_init_stack(&cl); SET_BDEV_STATE(&dc->sb, BDEV_STATE_STALE); @@ -868,46 +1074,91 @@ void bch_cached_dev_run(struct cached_dev *dc) closure_sync(&cl); } - add_disk(d->disk); + ret = add_disk(d->disk); + if (ret) + goto out; bd_link_disk_holder(dc->bdev, dc->disk.disk); - /* won't show up in the uevent file, use udevadm monitor -e instead - * only class / kset properties are persistent */ + /* + * won't show up in the uevent file, use udevadm monitor -e instead + * only class / kset properties are persistent + */ kobject_uevent_env(&disk_to_dev(d->disk)->kobj, KOBJ_CHANGE, env); + + if (sysfs_create_link(&d->kobj, &disk_to_dev(d->disk)->kobj, "dev") || + sysfs_create_link(&disk_to_dev(d->disk)->kobj, + &d->kobj, "bcache")) { + pr_err("Couldn't create bcache dev <-> disk sysfs symlinks\n"); + ret = -ENOMEM; + goto out; + } + + dc->status_update_thread = kthread_run(cached_dev_status_update, + dc, "bcache_status_update"); + if (IS_ERR(dc->status_update_thread)) { + pr_warn("failed to create bcache_status_update kthread, continue to run without monitoring backing device status\n"); + } + +out: kfree(env[1]); kfree(env[2]); + kfree(buf); + return ret; +} - if (sysfs_create_link(&d->kobj, &disk_to_dev(d->disk)->kobj, "dev") || - sysfs_create_link(&disk_to_dev(d->disk)->kobj, &d->kobj, "bcache")) - pr_debug("error creating sysfs link"); +/* + * If BCACHE_DEV_RATE_DW_RUNNING is set, it means routine of the delayed + * work dc->writeback_rate_update is running. Wait until the routine + * quits (BCACHE_DEV_RATE_DW_RUNNING is clear), then continue to + * cancel it. If BCACHE_DEV_RATE_DW_RUNNING is not clear after time_out + * seconds, give up waiting here and continue to cancel it too. + */ +static void cancel_writeback_rate_update_dwork(struct cached_dev *dc) +{ + int time_out = WRITEBACK_RATE_UPDATE_SECS_MAX * HZ; + + do { + if (!test_bit(BCACHE_DEV_RATE_DW_RUNNING, + &dc->disk.flags)) + break; + time_out--; + schedule_timeout_interruptible(1); + } while (time_out > 0); + + if (time_out == 0) + pr_warn("give up waiting for dc->writeback_write_update to quit\n"); + + cancel_delayed_work_sync(&dc->writeback_rate_update); } static void cached_dev_detach_finish(struct work_struct *w) { struct cached_dev *dc = container_of(w, struct cached_dev, detach); - char buf[BDEVNAME_SIZE]; - struct closure cl; - closure_init_stack(&cl); + struct cache_set *c = dc->disk.c; BUG_ON(!test_bit(BCACHE_DEV_DETACHING, &dc->disk.flags)); - BUG_ON(atomic_read(&dc->count)); + BUG_ON(refcount_read(&dc->count)); - mutex_lock(&bch_register_lock); - memset(&dc->sb.set_uuid, 0, 16); - SET_BDEV_STATE(&dc->sb, BDEV_STATE_NONE); + if (test_and_clear_bit(BCACHE_DEV_WB_RUNNING, &dc->disk.flags)) + cancel_writeback_rate_update_dwork(dc); - bch_write_bdev_super(dc, &cl); - closure_sync(&cl); + if (!IS_ERR_OR_NULL(dc->writeback_thread)) { + kthread_stop(dc->writeback_thread); + dc->writeback_thread = NULL; + } + + mutex_lock(&bch_register_lock); bcache_device_detach(&dc->disk); list_move(&dc->list, &uncached_devices); + calc_cached_dev_sectors(c); clear_bit(BCACHE_DEV_DETACHING, &dc->disk.flags); clear_bit(BCACHE_DEV_UNLINK_DONE, &dc->disk.flags); mutex_unlock(&bch_register_lock); - pr_info("Caching disabled for %s", bdevname(dc->bdev, buf)); + pr_info("Caching disabled for %pg\n", dc->bdev); /* Drop ref we took in cached_dev_detach() */ closure_put(&dc->disk.cl); @@ -930,66 +1181,80 @@ void bch_cached_dev_detach(struct cached_dev *dc) closure_get(&dc->disk.cl); bch_writeback_queue(dc); + cached_dev_put(dc); } -int bch_cached_dev_attach(struct cached_dev *dc, struct cache_set *c) +int bch_cached_dev_attach(struct cached_dev *dc, struct cache_set *c, + uint8_t *set_uuid) { - uint32_t rtime = cpu_to_le32(get_seconds()); + uint32_t rtime = cpu_to_le32((u32)ktime_get_real_seconds()); struct uuid_entry *u; - char buf[BDEVNAME_SIZE]; - - bdevname(dc->bdev, buf); + struct cached_dev *exist_dc, *t; + int ret = 0; - if (memcmp(dc->sb.set_uuid, c->sb.set_uuid, 16)) + if ((set_uuid && memcmp(set_uuid, c->set_uuid, 16)) || + (!set_uuid && memcmp(dc->sb.set_uuid, c->set_uuid, 16))) return -ENOENT; if (dc->disk.c) { - pr_err("Can't attach %s: already attached", buf); + pr_err("Can't attach %pg: already attached\n", dc->bdev); return -EINVAL; } if (test_bit(CACHE_SET_STOPPING, &c->flags)) { - pr_err("Can't attach %s: shutting down", buf); + pr_err("Can't attach %pg: shutting down\n", dc->bdev); return -EINVAL; } - if (dc->sb.block_size < c->sb.block_size) { + if (dc->sb.block_size < c->cache->sb.block_size) { /* Will die */ - pr_err("Couldn't attach %s: block size less than set's block size", - buf); + pr_err("Couldn't attach %pg: block size less than set's block size\n", + dc->bdev); return -EINVAL; } + /* Check whether already attached */ + list_for_each_entry_safe(exist_dc, t, &c->cached_devs, list) { + if (!memcmp(dc->sb.uuid, exist_dc->sb.uuid, 16)) { + pr_err("Tried to attach %pg but duplicate UUID already attached\n", + dc->bdev); + + return -EINVAL; + } + } + u = uuid_find(c, dc->sb.uuid); if (u && (BDEV_STATE(&dc->sb) == BDEV_STATE_STALE || BDEV_STATE(&dc->sb) == BDEV_STATE_NONE)) { memcpy(u->uuid, invalid_uuid, 16); - u->invalidated = cpu_to_le32(get_seconds()); + u->invalidated = cpu_to_le32((u32)ktime_get_real_seconds()); u = NULL; } if (!u) { if (BDEV_STATE(&dc->sb) == BDEV_STATE_DIRTY) { - pr_err("Couldn't find uuid for %s in set", buf); + pr_err("Couldn't find uuid for %pg in set\n", dc->bdev); return -ENOENT; } u = uuid_find_empty(c); if (!u) { - pr_err("Not caching %s, no room for UUID", buf); + pr_err("Not caching %pg, no room for UUID\n", dc->bdev); return -EINVAL; } } - /* Deadlocks since we're called via sysfs... - sysfs_remove_file(&dc->kobj, &sysfs_attach); + /* + * Deadlocks since we're called via sysfs... + * sysfs_remove_file(&dc->kobj, &sysfs_attach); */ if (bch_is_zero(u->uuid, 16)) { struct closure cl; + closure_init_stack(&cl); memcpy(u->uuid, dc->sb.uuid, 16); @@ -997,7 +1262,7 @@ int bch_cached_dev_attach(struct cached_dev *dc, struct cache_set *c) u->first_reg = u->last_reg = rtime; bch_uuid_write(c); - memcpy(dc->sb.set_uuid, c->sb.set_uuid, 16); + memcpy(dc->sb.set_uuid, c->set_uuid, 16); SET_BDEV_STATE(&dc->sb, BDEV_STATE_CLEAN); bch_write_bdev_super(dc, &cl); @@ -1011,39 +1276,64 @@ int bch_cached_dev_attach(struct cached_dev *dc, struct cache_set *c) list_move(&dc->list, &c->cached_devs); calc_cached_dev_sectors(c); - smp_wmb(); /* * dc->c must be set before dc->count != 0 - paired with the mb in * cached_dev_get() */ - atomic_set(&dc->count, 1); + smp_wmb(); + refcount_set(&dc->count, 1); /* Block writeback thread, but spawn it */ down_write(&dc->writeback_lock); if (bch_cached_dev_writeback_start(dc)) { up_write(&dc->writeback_lock); + pr_err("Couldn't start writeback facilities for %s\n", + dc->disk.disk->disk_name); return -ENOMEM; } if (BDEV_STATE(&dc->sb) == BDEV_STATE_DIRTY) { - bch_sectors_dirty_init(dc); atomic_set(&dc->has_dirty, 1); - atomic_inc(&dc->count); bch_writeback_queue(dc); } - bch_cached_dev_run(dc); + bch_sectors_dirty_init(&dc->disk); + + ret = bch_cached_dev_run(dc); + if (ret && (ret != -EBUSY)) { + up_write(&dc->writeback_lock); + /* + * bch_register_lock is held, bcache_device_stop() is not + * able to be directly called. The kthread and kworker + * created previously in bch_cached_dev_writeback_start() + * have to be stopped manually here. + */ + kthread_stop(dc->writeback_thread); + cancel_writeback_rate_update_dwork(dc); + pr_err("Couldn't run cached device %pg\n", dc->bdev); + return ret; + } + bcache_device_link(&dc->disk, c, "bdev"); + atomic_inc(&c->attached_dev_nr); + + if (bch_has_feature_obso_large_bucket(&(c->cache->sb))) { + pr_err("The obsoleted large bucket layout is unsupported, set the bcache device into read-only\n"); + pr_err("Please update to the latest bcache-tools to create the cache device\n"); + set_disk_ro(dc->disk.disk, 1); + } /* Allow the writeback thread to proceed */ up_write(&dc->writeback_lock); - pr_info("Caching %s as %s on set %pU", - bdevname(dc->bdev, buf), dc->disk.disk->disk_name, - dc->disk.c->sb.set_uuid); + pr_info("Caching %pg as %s on set %pU\n", + dc->bdev, + dc->disk.disk->disk_name, + dc->disk.c->set_uuid); return 0; } +/* when dc->disk.kobj released */ void bch_cached_dev_release(struct kobject *kobj) { struct cached_dev *dc = container_of(kobj, struct cached_dev, @@ -1052,34 +1342,43 @@ void bch_cached_dev_release(struct kobject *kobj) module_put(THIS_MODULE); } -static void cached_dev_free(struct closure *cl) +static CLOSURE_CALLBACK(cached_dev_free) { - struct cached_dev *dc = container_of(cl, struct cached_dev, disk.cl); + closure_type(dc, struct cached_dev, disk.cl); + + if (test_and_clear_bit(BCACHE_DEV_WB_RUNNING, &dc->disk.flags)) + cancel_writeback_rate_update_dwork(dc); - cancel_delayed_work_sync(&dc->writeback_rate_update); if (!IS_ERR_OR_NULL(dc->writeback_thread)) kthread_stop(dc->writeback_thread); + if (!IS_ERR_OR_NULL(dc->status_update_thread)) + kthread_stop(dc->status_update_thread); mutex_lock(&bch_register_lock); - if (atomic_read(&dc->running)) + if (atomic_read(&dc->running)) { bd_unlink_disk_holder(dc->bdev, dc->disk.disk); + del_gendisk(dc->disk.disk); + } bcache_device_free(&dc->disk); list_del(&dc->list); mutex_unlock(&bch_register_lock); - if (!IS_ERR_OR_NULL(dc->bdev)) - blkdev_put(dc->bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL); + if (dc->sb_disk) + folio_put(virt_to_folio(dc->sb_disk)); + + if (dc->bdev_file) + fput(dc->bdev_file); wake_up(&unregister_wait); kobject_put(&dc->disk.kobj); } -static void cached_dev_flush(struct closure *cl) +static CLOSURE_CALLBACK(cached_dev_flush) { - struct cached_dev *dc = container_of(cl, struct cached_dev, disk.cl); + closure_type(dc, struct cached_dev, disk.cl); struct bcache_device *d = &dc->disk; mutex_lock(&bch_register_lock); @@ -1089,10 +1388,10 @@ static void cached_dev_flush(struct closure *cl) bch_cache_accounting_destroy(&dc->accounting); kobject_del(&d->kobj); - continue_at(cl, cached_dev_free, system_wq); + continue_at(cl, cached_dev_free, system_percpu_wq); } -static int cached_dev_init(struct cached_dev *dc, unsigned block_size) +static int cached_dev_init(struct cached_dev *dc, unsigned int block_size) { int ret; struct io *io; @@ -1101,7 +1400,7 @@ static int cached_dev_init(struct cached_dev *dc, unsigned block_size) __module_get(THIS_MODULE); INIT_LIST_HEAD(&dc->list); closure_init(&dc->disk.cl, NULL); - set_closure_fn(&dc->disk.cl, cached_dev_flush, system_wq); + set_closure_fn(&dc->disk.cl, cached_dev_flush, system_percpu_wq); kobject_init(&dc->disk.kobj, &bch_cached_dev_ktype); INIT_WORK(&dc->detach, cached_dev_detach_finish); sema_init(&dc->sb_write_mutex, 1); @@ -1116,23 +1415,21 @@ static int cached_dev_init(struct cached_dev *dc, unsigned block_size) hlist_add_head(&io->hash, dc->io_hash + RECENT_IO); } - dc->disk.stripe_size = q->limits.io_opt >> 9; - - if (dc->disk.stripe_size) - dc->partial_stripes_expensive = - q->limits.raid_partial_stripes_expensive; + if (bdev_io_opt(dc->bdev)) + dc->partial_stripes_expensive = !!(q->limits.features & + BLK_FEAT_RAID_PARTIAL_STRIPES_EXPENSIVE); ret = bcache_device_init(&dc->disk, block_size, - dc->bdev->bd_part->nr_sects - dc->sb.data_offset); + bdev_nr_sectors(dc->bdev) - dc->sb.data_offset, + dc->bdev, &bcache_cached_ops); if (ret) return ret; - set_capacity(dc->disk.disk, - dc->bdev->bd_part->nr_sects - dc->sb.data_offset); - - dc->disk.disk->queue->backing_dev_info->ra_pages = - max(dc->disk.disk->queue->backing_dev_info->ra_pages, - q->backing_dev_info->ra_pages); + atomic_set(&dc->io_errors, 0); + dc->io_disable = false; + dc->error_limit = DEFAULT_CACHED_DEV_ERROR_LIMIT; + /* default to auto */ + dc->stop_when_cache_set_failed = BCH_CACHED_DEV_STOP_AUTO; bch_cached_dev_request_init(dc); bch_cached_dev_writeback_init(dc); @@ -1141,50 +1438,53 @@ static int cached_dev_init(struct cached_dev *dc, unsigned block_size) /* Cached device - bcache superblock */ -static void register_bdev(struct cache_sb *sb, struct page *sb_page, - struct block_device *bdev, +static int register_bdev(struct cache_sb *sb, struct cache_sb_disk *sb_disk, + struct file *bdev_file, struct cached_dev *dc) { - char name[BDEVNAME_SIZE]; const char *err = "cannot allocate memory"; struct cache_set *c; + int ret = -ENOMEM; memcpy(&dc->sb, sb, sizeof(struct cache_sb)); - dc->bdev = bdev; - dc->bdev->bd_holder = dc; - - bio_init(&dc->sb_bio, dc->sb_bio.bi_inline_vecs, 1); - dc->sb_bio.bi_io_vec[0].bv_page = sb_page; - get_page(sb_page); + dc->bdev_file = bdev_file; + dc->bdev = file_bdev(bdev_file); + dc->sb_disk = sb_disk; if (cached_dev_init(dc, sb->block_size << 9)) goto err; err = "error creating kobject"; - if (kobject_add(&dc->disk.kobj, &part_to_dev(bdev->bd_part)->kobj, - "bcache")) + if (kobject_add(&dc->disk.kobj, bdev_kobj(dc->bdev), "bcache")) goto err; if (bch_cache_accounting_add_kobjs(&dc->accounting, &dc->disk.kobj)) goto err; - pr_info("registered backing device %s", bdevname(bdev, name)); + pr_info("registered backing device %pg\n", dc->bdev); list_add(&dc->list, &uncached_devices); + /* attach to a matched cache set if it exists */ list_for_each_entry(c, &bch_cache_sets, list) - bch_cached_dev_attach(dc, c); + bch_cached_dev_attach(dc, c, NULL); if (BDEV_STATE(&dc->sb) == BDEV_STATE_NONE || - BDEV_STATE(&dc->sb) == BDEV_STATE_STALE) - bch_cached_dev_run(dc); + BDEV_STATE(&dc->sb) == BDEV_STATE_STALE) { + err = "failed to run cached device"; + ret = bch_cached_dev_run(dc); + if (ret) + goto err; + } - return; + return 0; err: - pr_notice("error opening %s: %s", bdevname(bdev, name), err); + pr_notice("error %pg: %s\n", dc->bdev, err); bcache_device_stop(&dc->disk); + return ret; } /* Flash only volumes */ +/* When d->kobj released */ void bch_flash_dev_release(struct kobject *kobj) { struct bcache_device *d = container_of(kobj, struct bcache_device, @@ -1192,54 +1492,71 @@ void bch_flash_dev_release(struct kobject *kobj) kfree(d); } -static void flash_dev_free(struct closure *cl) +static CLOSURE_CALLBACK(flash_dev_free) { - struct bcache_device *d = container_of(cl, struct bcache_device, cl); + closure_type(d, struct bcache_device, cl); + mutex_lock(&bch_register_lock); + atomic_long_sub(bcache_dev_sectors_dirty(d), + &d->c->flash_dev_dirty_sectors); + del_gendisk(d->disk); bcache_device_free(d); mutex_unlock(&bch_register_lock); kobject_put(&d->kobj); } -static void flash_dev_flush(struct closure *cl) +static CLOSURE_CALLBACK(flash_dev_flush) { - struct bcache_device *d = container_of(cl, struct bcache_device, cl); + closure_type(d, struct bcache_device, cl); mutex_lock(&bch_register_lock); bcache_device_unlink(d); mutex_unlock(&bch_register_lock); kobject_del(&d->kobj); - continue_at(cl, flash_dev_free, system_wq); + continue_at(cl, flash_dev_free, system_percpu_wq); } static int flash_dev_run(struct cache_set *c, struct uuid_entry *u) { + int err = -ENOMEM; struct bcache_device *d = kzalloc(sizeof(struct bcache_device), GFP_KERNEL); if (!d) - return -ENOMEM; + goto err_ret; closure_init(&d->cl, NULL); - set_closure_fn(&d->cl, flash_dev_flush, system_wq); + set_closure_fn(&d->cl, flash_dev_flush, system_percpu_wq); kobject_init(&d->kobj, &bch_flash_dev_ktype); - if (bcache_device_init(d, block_bytes(c), u->sectors)) + if (bcache_device_init(d, block_bytes(c->cache), u->sectors, + NULL, &bcache_flash_ops)) goto err; bcache_device_attach(d, c, u - c->uuids); + bch_sectors_dirty_init(d); bch_flash_dev_request_init(d); - add_disk(d->disk); + err = add_disk(d->disk); + if (err) + goto err; - if (kobject_add(&d->kobj, &disk_to_dev(d->disk)->kobj, "bcache")) + err = kobject_add(&d->kobj, &disk_to_dev(d->disk)->kobj, "bcache"); + if (err) goto err; bcache_device_link(d, c, "volume"); + if (bch_has_feature_obso_large_bucket(&c->cache->sb)) { + pr_err("The obsoleted large bucket layout is unsupported, set the bcache device into read-only\n"); + pr_err("Please update to the latest bcache-tools to create the cache device\n"); + set_disk_ro(d->disk, 1); + } + return 0; err: kobject_put(&d->kobj); - return -ENOMEM; +err_ret: + return err; } static int flash_devs_run(struct cache_set *c) @@ -1268,13 +1585,13 @@ int bch_flash_dev_create(struct cache_set *c, uint64_t size) u = uuid_find_empty(c); if (!u) { - pr_err("Can't create volume, no room for UUID"); + pr_err("Can't create volume, no room for UUID\n"); return -EINVAL; } get_random_bytes(u->uuid, 16); memset(u->label, 0, 32); - u->first_reg = u->last_reg = cpu_to_le32(get_seconds()); + u->first_reg = u->last_reg = cpu_to_le32((u32)ktime_get_real_seconds()); SET_UUID_FLASH_ONLY(u, 1); u->sectors = size >> 9; @@ -1284,28 +1601,51 @@ int bch_flash_dev_create(struct cache_set *c, uint64_t size) return flash_dev_run(c, u); } +bool bch_cached_dev_error(struct cached_dev *dc) +{ + if (!dc || test_bit(BCACHE_DEV_CLOSING, &dc->disk.flags)) + return false; + + dc->io_disable = true; + /* make others know io_disable is true earlier */ + smp_mb(); + + pr_err("stop %s: too many IO errors on backing device %pg\n", + dc->disk.disk->disk_name, dc->bdev); + + bcache_device_stop(&dc->disk); + return true; +} + /* Cache set */ __printf(2, 3) bool bch_cache_set_error(struct cache_set *c, const char *fmt, ...) { + struct va_format vaf; va_list args; if (c->on_error != ON_ERROR_PANIC && test_bit(CACHE_SET_STOPPING, &c->flags)) return false; - /* XXX: we can be called from atomic context - acquire_console_sem(); - */ + if (test_and_set_bit(CACHE_SET_IO_DISABLE, &c->flags)) + pr_info("CACHE_SET_IO_DISABLE already set\n"); - printk(KERN_ERR "bcache: error on %pU: ", c->sb.set_uuid); + /* + * XXX: we can be called from atomic context + * acquire_console_sem(); + */ va_start(args, fmt); - vprintk(fmt, args); - va_end(args); - printk(", disabling caching\n"); + vaf.fmt = fmt; + vaf.va = &args; + + pr_err("error on %pU: %pV, disabling caching\n", + c->set_uuid, &vaf); + + va_end(args); if (c->on_error == ON_ERROR_PANIC) panic("panic forced after error\n"); @@ -1314,91 +1654,92 @@ bool bch_cache_set_error(struct cache_set *c, const char *fmt, ...) return true; } +/* When c->kobj released */ void bch_cache_set_release(struct kobject *kobj) { struct cache_set *c = container_of(kobj, struct cache_set, kobj); + kfree(c); module_put(THIS_MODULE); } -static void cache_set_free(struct closure *cl) +static CLOSURE_CALLBACK(cache_set_free) { - struct cache_set *c = container_of(cl, struct cache_set, cl); + closure_type(c, struct cache_set, cl); struct cache *ca; - unsigned i; - if (!IS_ERR_OR_NULL(c->debug)) - debugfs_remove(c->debug); + debugfs_remove(c->debug); bch_open_buckets_free(c); bch_btree_cache_free(c); bch_journal_free(c); - for_each_cache(ca, c, i) - if (ca) { - ca->set = NULL; - c->cache[ca->sb.nr_this_dev] = NULL; - kobject_put(&ca->kobj); - } - + mutex_lock(&bch_register_lock); bch_bset_sort_state_free(&c->sort); - free_pages((unsigned long) c->uuids, ilog2(bucket_pages(c))); + free_pages((unsigned long) c->uuids, ilog2(meta_bucket_pages(&c->cache->sb))); + + ca = c->cache; + if (ca) { + ca->set = NULL; + c->cache = NULL; + kobject_put(&ca->kobj); + } + if (c->moving_gc_wq) destroy_workqueue(c->moving_gc_wq); - if (c->bio_split) - bioset_free(c->bio_split); - if (c->fill_iter) - mempool_destroy(c->fill_iter); - if (c->bio_meta) - mempool_destroy(c->bio_meta); - if (c->search) - mempool_destroy(c->search); + bioset_exit(&c->bio_split); + mempool_exit(&c->fill_iter); + mempool_exit(&c->bio_meta); + mempool_exit(&c->search); kfree(c->devices); - mutex_lock(&bch_register_lock); list_del(&c->list); mutex_unlock(&bch_register_lock); - pr_info("Cache set %pU unregistered", c->sb.set_uuid); + pr_info("Cache set %pU unregistered\n", c->set_uuid); wake_up(&unregister_wait); closure_debug_destroy(&c->cl); kobject_put(&c->kobj); } -static void cache_set_flush(struct closure *cl) +static CLOSURE_CALLBACK(cache_set_flush) { - struct cache_set *c = container_of(cl, struct cache_set, caching); - struct cache *ca; + closure_type(c, struct cache_set, caching); + struct cache *ca = c->cache; struct btree *b; - unsigned i; - - if (!c) - closure_return(cl); bch_cache_accounting_destroy(&c->accounting); kobject_put(&c->internal); kobject_del(&c->kobj); - if (c->gc_thread) + if (!IS_ERR_OR_NULL(c->gc_thread)) kthread_stop(c->gc_thread); if (!IS_ERR_OR_NULL(c->root)) list_add(&c->root->list, &c->btree_cache); - /* Should skip this if we're unregistering because of an error */ - list_for_each_entry(b, &c->btree_cache, list) { - mutex_lock(&b->write_lock); - if (btree_node_dirty(b)) - __bch_btree_node_write(b, NULL); - mutex_unlock(&b->write_lock); - } + /* + * Avoid flushing cached nodes if cache set is retiring + * due to too many I/O errors detected. + */ + if (!test_bit(CACHE_SET_IO_DISABLE, &c->flags)) + list_for_each_entry(b, &c->btree_cache, list) { + mutex_lock(&b->write_lock); + if (btree_node_dirty(b)) + __bch_btree_node_write(b, NULL); + mutex_unlock(&b->write_lock); + } - for_each_cache(ca, c, i) - if (ca->alloc_thread) - kthread_stop(ca->alloc_thread); + /* + * If the register_cache_set() call to bch_cache_set_alloc() failed, + * ca has not been assigned a value and return error. + * So we need check ca is not NULL during bch_cache_set_unregister(). + */ + if (ca && ca->alloc_thread) + kthread_stop(ca->alloc_thread); if (c->journal.cur) { cancel_delayed_work_sync(&c->journal.work); @@ -1409,34 +1750,96 @@ static void cache_set_flush(struct closure *cl) closure_return(cl); } -static void __cache_set_unregister(struct closure *cl) +/* + * This function is only called when CACHE_SET_IO_DISABLE is set, which means + * cache set is unregistering due to too many I/O errors. In this condition, + * the bcache device might be stopped, it depends on stop_when_cache_set_failed + * value and whether the broken cache has dirty data: + * + * dc->stop_when_cache_set_failed dc->has_dirty stop bcache device + * BCH_CACHED_STOP_AUTO 0 NO + * BCH_CACHED_STOP_AUTO 1 YES + * BCH_CACHED_DEV_STOP_ALWAYS 0 YES + * BCH_CACHED_DEV_STOP_ALWAYS 1 YES + * + * The expected behavior is, if stop_when_cache_set_failed is configured to + * "auto" via sysfs interface, the bcache device will not be stopped if the + * backing device is clean on the broken cache device. + */ +static void conditional_stop_bcache_device(struct cache_set *c, + struct bcache_device *d, + struct cached_dev *dc) +{ + if (dc->stop_when_cache_set_failed == BCH_CACHED_DEV_STOP_ALWAYS) { + pr_warn("stop_when_cache_set_failed of %s is \"always\", stop it for failed cache set %pU.\n", + d->disk->disk_name, c->set_uuid); + bcache_device_stop(d); + } else if (atomic_read(&dc->has_dirty)) { + /* + * dc->stop_when_cache_set_failed == BCH_CACHED_STOP_AUTO + * and dc->has_dirty == 1 + */ + pr_warn("stop_when_cache_set_failed of %s is \"auto\" and cache is dirty, stop it to avoid potential data corruption.\n", + d->disk->disk_name); + /* + * There might be a small time gap that cache set is + * released but bcache device is not. Inside this time + * gap, regular I/O requests will directly go into + * backing device as no cache set attached to. This + * behavior may also introduce potential inconsistence + * data in writeback mode while cache is dirty. + * Therefore before calling bcache_device_stop() due + * to a broken cache device, dc->io_disable should be + * explicitly set to true. + */ + dc->io_disable = true; + /* make others know io_disable is true earlier */ + smp_mb(); + bcache_device_stop(d); + } else { + /* + * dc->stop_when_cache_set_failed == BCH_CACHED_STOP_AUTO + * and dc->has_dirty == 0 + */ + pr_warn("stop_when_cache_set_failed of %s is \"auto\" and cache is clean, keep it alive.\n", + d->disk->disk_name); + } +} + +static CLOSURE_CALLBACK(__cache_set_unregister) { - struct cache_set *c = container_of(cl, struct cache_set, caching); + closure_type(c, struct cache_set, caching); struct cached_dev *dc; + struct bcache_device *d; size_t i; mutex_lock(&bch_register_lock); - for (i = 0; i < c->nr_uuids; i++) - if (c->devices[i]) { - if (!UUID_FLASH_ONLY(&c->uuids[i]) && - test_bit(CACHE_SET_UNREGISTERING, &c->flags)) { - dc = container_of(c->devices[i], - struct cached_dev, disk); - bch_cached_dev_detach(dc); - } else { - bcache_device_stop(c->devices[i]); - } + for (i = 0; i < c->devices_max_used; i++) { + d = c->devices[i]; + if (!d) + continue; + + if (!UUID_FLASH_ONLY(&c->uuids[i]) && + test_bit(CACHE_SET_UNREGISTERING, &c->flags)) { + dc = container_of(d, struct cached_dev, disk); + bch_cached_dev_detach(dc); + if (test_bit(CACHE_SET_IO_DISABLE, &c->flags)) + conditional_stop_bcache_device(c, d, dc); + } else { + bcache_device_stop(d); } + } mutex_unlock(&bch_register_lock); - continue_at(cl, cache_set_flush, system_wq); + continue_at(cl, cache_set_flush, system_percpu_wq); } void bch_cache_set_stop(struct cache_set *c) { if (!test_and_set_bit(CACHE_SET_STOPPING, &c->flags)) + /* closure_fn set to __cache_set_unregister() */ closure_queue(&c->caching); } @@ -1446,22 +1849,24 @@ void bch_cache_set_unregister(struct cache_set *c) bch_cache_set_stop(c); } -#define alloc_bucket_pages(gfp, c) \ - ((void *) __get_free_pages(__GFP_ZERO|gfp, ilog2(bucket_pages(c)))) +#define alloc_meta_bucket_pages(gfp, sb) \ + ((void *) __get_free_pages(__GFP_ZERO|__GFP_COMP|gfp, ilog2(meta_bucket_pages(sb)))) struct cache_set *bch_cache_set_alloc(struct cache_sb *sb) { int iter_size; + struct cache *ca = container_of(sb, struct cache, sb); struct cache_set *c = kzalloc(sizeof(struct cache_set), GFP_KERNEL); + if (!c) return NULL; __module_get(THIS_MODULE); closure_init(&c->cl, NULL); - set_closure_fn(&c->cl, cache_set_free, system_wq); + set_closure_fn(&c->cl, cache_set_free, system_percpu_wq); closure_init(&c->caching, &c->cl); - set_closure_fn(&c->caching, __cache_set_unregister, system_wq); + set_closure_fn(&c->caching, __cache_set_unregister, system_percpu_wq); /* Maybe create continue_at_noreturn() and use it here? */ closure_set_stopped(&c->cl); @@ -1472,16 +1877,16 @@ struct cache_set *bch_cache_set_alloc(struct cache_sb *sb) bch_cache_accounting_init(&c->accounting, &c->cl); - memcpy(c->sb.set_uuid, sb->set_uuid, 16); - c->sb.block_size = sb->block_size; - c->sb.bucket_size = sb->bucket_size; - c->sb.nr_in_set = sb->nr_in_set; - c->sb.last_mount = sb->last_mount; + memcpy(c->set_uuid, sb->set_uuid, 16); + + c->cache = ca; + c->cache->set = c; c->bucket_bits = ilog2(sb->bucket_size); c->block_bits = ilog2(sb->block_size); - c->nr_uuids = bucket_bytes(c) / sizeof(struct uuid_entry); - - c->btree_pages = bucket_pages(c); + c->nr_uuids = meta_bucket_bytes(sb) / sizeof(struct uuid_entry); + c->devices_max_used = 0; + atomic_set(&c->attached_dev_nr, 0); + c->btree_pages = meta_bucket_pages(sb); if (c->btree_pages > BTREE_MAX_PAGES) c->btree_pages = max_t(int, c->btree_pages / 4, BTREE_MAX_PAGES); @@ -1489,6 +1894,7 @@ struct cache_set *bch_cache_set_alloc(struct cache_sb *sb) sema_init(&c->sb_write_mutex, 1); mutex_init(&c->bucket_lock); init_waitqueue_head(&c->btree_cache_wait); + spin_lock_init(&c->btree_cannibalize_lock); init_waitqueue_head(&c->bucket_wait); init_waitqueue_head(&c->gc_wait); sema_init(&c->uuid_write_mutex, 1); @@ -1506,33 +1912,55 @@ struct cache_set *bch_cache_set_alloc(struct cache_sb *sb) INIT_LIST_HEAD(&c->btree_cache_freed); INIT_LIST_HEAD(&c->data_buckets); - c->search = mempool_create_slab_pool(32, bch_search_cache); - if (!c->search) + iter_size = sizeof(struct btree_iter) + + ((meta_bucket_pages(sb) * PAGE_SECTORS) / sb->block_size) * + sizeof(struct btree_iter_set); + + c->devices = kcalloc(c->nr_uuids, sizeof(void *), GFP_KERNEL); + if (!c->devices) + goto err; + + if (mempool_init_slab_pool(&c->search, 32, bch_search_cache)) + goto err; + + if (mempool_init_kmalloc_pool(&c->bio_meta, 2, + sizeof(struct bbio) + + sizeof(struct bio_vec) * meta_bucket_pages(sb))) + goto err; + + if (mempool_init_kmalloc_pool(&c->fill_iter, 1, iter_size)) + goto err; + + if (bioset_init(&c->bio_split, 4, offsetof(struct bbio, bio), + BIOSET_NEED_RESCUER)) goto err; - iter_size = (sb->bucket_size / sb->block_size + 1) * - sizeof(struct btree_iter_set); - - if (!(c->devices = kzalloc(c->nr_uuids * sizeof(void *), GFP_KERNEL)) || - !(c->bio_meta = mempool_create_kmalloc_pool(2, - sizeof(struct bbio) + sizeof(struct bio_vec) * - bucket_pages(c))) || - !(c->fill_iter = mempool_create_kmalloc_pool(1, iter_size)) || - !(c->bio_split = bioset_create(4, offsetof(struct bbio, bio), - BIOSET_NEED_BVECS | - BIOSET_NEED_RESCUER)) || - !(c->uuids = alloc_bucket_pages(GFP_KERNEL, c)) || - !(c->moving_gc_wq = alloc_workqueue("bcache_gc", - WQ_MEM_RECLAIM, 0)) || - bch_journal_alloc(c) || - bch_btree_cache_alloc(c) || - bch_open_buckets_alloc(c) || - bch_bset_sort_state_init(&c->sort, ilog2(c->btree_pages))) + c->uuids = alloc_meta_bucket_pages(GFP_KERNEL, sb); + if (!c->uuids) + goto err; + + c->moving_gc_wq = alloc_workqueue("bcache_gc", + WQ_MEM_RECLAIM | WQ_PERCPU, 0); + if (!c->moving_gc_wq) + goto err; + + if (bch_journal_alloc(c)) + goto err; + + if (bch_btree_cache_alloc(c)) + goto err; + + if (bch_open_buckets_alloc(c)) + goto err; + + if (bch_bset_sort_state_init(&c->sort, ilog2(c->btree_pages))) goto err; c->congested_read_threshold_us = 2000; c->congested_write_threshold_us = 20000; - c->error_limit = 8 << IO_ERROR_SHIFT; + c->error_limit = DEFAULT_IO_ERROR_LIMIT; + c->idle_max_writeback_rate_enabled = 1; + WARN_ON(test_and_clear_bit(CACHE_SET_IO_DISABLE, &c->flags)); return c; err: @@ -1540,22 +1968,21 @@ err: return NULL; } -static void run_cache_set(struct cache_set *c) +static int run_cache_set(struct cache_set *c) { const char *err = "cannot allocate memory"; struct cached_dev *dc, *t; - struct cache *ca; + struct cache *ca = c->cache; struct closure cl; - unsigned i; + LIST_HEAD(journal); + struct journal_replay *l; closure_init_stack(&cl); - for_each_cache(ca, c, i) - c->nbuckets += ca->sb.nbuckets; + c->nbuckets = ca->sb.nbuckets; set_gc_sectors(c); - if (CACHE_SYNC(&c->sb)) { - LIST_HEAD(journal); + if (CACHE_SYNC(&c->cache->sb)) { struct bkey *k; struct jset *j; @@ -1563,7 +1990,7 @@ static void run_cache_set(struct cache_set *c) if (bch_journal_read(c, &journal)) goto err; - pr_debug("btree_journal_read() done"); + pr_debug("btree_journal_read() done\n"); err = "no journal entries found"; if (list_empty(&journal)) @@ -1572,8 +1999,8 @@ static void run_cache_set(struct cache_set *c) j = &list_entry(journal.prev, struct journal_replay, list)->j; err = "IO error reading priorities"; - for_each_cache(ca, c, i) - prio_read(ca, j->prio_bucket[ca->sb.nr_this_dev]); + if (prio_read(ca, j->prio_bucket[ca->sb.nr_this_dev])) + goto err; /* * If prio_read() fails it'll call cache_set_error and we'll @@ -1588,8 +2015,10 @@ static void run_cache_set(struct cache_set *c) goto err; err = "error reading btree root"; - c->root = bch_btree_node_get(c, NULL, k, j->btree_level, true, NULL); - if (IS_ERR_OR_NULL(c->root)) + c->root = bch_btree_node_get(c, NULL, k, + j->btree_level, + true, NULL); + if (IS_ERR(c->root)) goto err; list_del_init(&c->root->list); @@ -1605,7 +2034,7 @@ static void run_cache_set(struct cache_set *c) bch_journal_mark(c, &journal); bch_initial_gc_finish(c); - pr_debug("btree_check() done"); + pr_debug("btree_check() done\n"); /* * bcache_journal_next() can't happen sooner, or @@ -1615,9 +2044,8 @@ static void run_cache_set(struct cache_set *c) bch_journal_next(&c->journal); err = "error starting allocator thread"; - for_each_cache(ca, c, i) - if (bch_cache_allocator_start(ca)) - goto err; + if (bch_cache_allocator_start(ca)) + goto err; /* * First place it's safe to allocate: btree_check() and @@ -1632,30 +2060,27 @@ static void run_cache_set(struct cache_set *c) if (j->version < BCACHE_JSET_VERSION_UUID) __uuid_write(c); - bch_journal_replay(c, &journal); + err = "bcache: replay journal failed"; + if (bch_journal_replay(c, &journal)) + goto err; } else { - pr_notice("invalidating existing data"); + unsigned int j; - for_each_cache(ca, c, i) { - unsigned j; + pr_notice("invalidating existing data\n"); + ca->sb.keys = clamp_t(int, ca->sb.nbuckets >> 7, + 2, SB_JOURNAL_BUCKETS); - ca->sb.keys = clamp_t(int, ca->sb.nbuckets >> 7, - 2, SB_JOURNAL_BUCKETS); - - for (j = 0; j < ca->sb.keys; j++) - ca->sb.d[j] = ca->sb.first_bucket + j; - } + for (j = 0; j < ca->sb.keys; j++) + ca->sb.d[j] = ca->sb.first_bucket + j; bch_initial_gc_finish(c); err = "error starting allocator thread"; - for_each_cache(ca, c, i) - if (bch_cache_allocator_start(ca)) - goto err; + if (bch_cache_allocator_start(ca)) + goto err; mutex_lock(&c->bucket_lock); - for_each_cache(ca, c, i) - bch_prio_write(ca); + bch_prio_write(ca, true); mutex_unlock(&c->bucket_lock); err = "cannot allocate new UUID bucket"; @@ -1664,7 +2089,7 @@ static void run_cache_set(struct cache_set *c) err = "cannot allocate new btree root"; c->root = __bch_btree_node_alloc(c, NULL, 0, true, NULL); - if (IS_ERR_OR_NULL(c->root)) + if (IS_ERR(c->root)) goto err; mutex_lock(&c->root->write_lock); @@ -1680,7 +2105,7 @@ static void run_cache_set(struct cache_set *c) * everything is set up - fortunately journal entries won't be * written until the SET_CACHE_SYNC() here: */ - SET_CACHE_SYNC(&c->sb, true); + SET_CACHE_SYNC(&c->cache->sb, true); bch_journal_next(&c->journal); bch_journal_meta(c, &cl); @@ -1691,27 +2116,32 @@ static void run_cache_set(struct cache_set *c) goto err; closure_sync(&cl); - c->sb.last_mount = get_seconds(); + c->cache->sb.last_mount = (u32)ktime_get_real_seconds(); bcache_write_super(c); + if (bch_has_feature_obso_large_bucket(&c->cache->sb)) + pr_err("Detect obsoleted large bucket layout, all attached bcache device will be read-only\n"); + list_for_each_entry_safe(dc, t, &uncached_devices, list) - bch_cached_dev_attach(dc, c); + bch_cached_dev_attach(dc, c, NULL); flash_devs_run(c); + bch_journal_space_reserve(&c->journal); set_bit(CACHE_SET_RUNNING, &c->flags); - return; + return 0; err: + while (!list_empty(&journal)) { + l = list_first_entry(&journal, struct journal_replay, list); + list_del(&l->list); + kfree(l); + } + closure_sync(&cl); - /* XXX: test this, it's broken */ + bch_cache_set_error(c, "%s", err); -} -static bool can_attach_cache(struct cache *ca, struct cache_set *c) -{ - return ca->sb.block_size == c->sb.block_size && - ca->sb.bucket_size == c->sb.bucket_size && - ca->sb.nr_in_set == c->sb.nr_in_set; + return -EIO; } static const char *register_cache_set(struct cache *ca) @@ -1721,16 +2151,10 @@ static const char *register_cache_set(struct cache *ca) struct cache_set *c; list_for_each_entry(c, &bch_cache_sets, list) - if (!memcmp(c->sb.set_uuid, ca->sb.set_uuid, 16)) { - if (c->cache[ca->sb.nr_this_dev]) + if (!memcmp(c->set_uuid, ca->sb.set_uuid, 16)) { + if (c->cache) return "duplicate cache set member"; - if (!can_attach_cache(ca, c)) - return "cache sb does not match set"; - - if (!CACHE_SYNC(&ca->sb)) - SET_CACHE_SYNC(&c->sb, false); - goto found; } @@ -1739,7 +2163,7 @@ static const char *register_cache_set(struct cache *ca) return err; err = "error creating kobject"; - if (kobject_add(&c->kobj, bcache_kobj, "%pU", c->sb.set_uuid) || + if (kobject_add(&c->kobj, bcache_kobj, "%pU", c->set_uuid) || kobject_add(&c->internal, &c->kobj, "internal")) goto err; @@ -1755,21 +2179,13 @@ found: sysfs_create_link(&c->kobj, &ca->kobj, buf)) goto err; - if (ca->sb.seq > c->sb.seq) { - c->sb.version = ca->sb.version; - memcpy(c->sb.set_uuid, ca->sb.set_uuid, 16); - c->sb.flags = ca->sb.flags; - c->sb.seq = ca->sb.seq; - pr_debug("set version = %llu", c->sb.version); - } - kobject_get(&ca->kobj); ca->set = c; - ca->set->cache[ca->sb.nr_this_dev] = ca; - c->cache_by_alloc[c->caches_loaded++] = ca; + ca->set->cache = ca; - if (c->caches_loaded == c->sb.nr_in_set) - run_cache_set(c); + err = "failed to run cache set"; + if (run_cache_set(c) < 0) + goto err; return NULL; err: @@ -1779,17 +2195,18 @@ err: /* Cache device */ +/* When ca->kobj released */ void bch_cache_release(struct kobject *kobj) { struct cache *ca = container_of(kobj, struct cache, kobj); - unsigned i; + unsigned int i; if (ca->set) { - BUG_ON(ca->set->cache[ca->sb.nr_this_dev] != ca); - ca->set->cache[ca->sb.nr_this_dev] = NULL; + BUG_ON(ca->set->cache != ca); + ca->set->cache = NULL; } - free_pages((unsigned long) ca->disk_buckets, ilog2(bucket_pages(ca))); + free_pages((unsigned long) ca->disk_buckets, ilog2(meta_bucket_pages(&ca->sb))); kfree(ca->prio_buckets); vfree(ca->buckets); @@ -1799,11 +2216,11 @@ void bch_cache_release(struct kobject *kobj) for (i = 0; i < RESERVE_NR; i++) free_fifo(&ca->free[i]); - if (ca->sb_bio.bi_inline_vecs[0].bv_page) - put_page(ca->sb_bio.bi_io_vec[0].bv_page); + if (ca->sb_disk) + folio_put(virt_to_folio(ca->sb_disk)); - if (!IS_ERR_OR_NULL(ca->bdev)) - blkdev_put(ca->bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL); + if (ca->bdev_file) + fput(ca->bdev_file); kfree(ca); module_put(THIS_MODULE); @@ -1812,65 +2229,182 @@ void bch_cache_release(struct kobject *kobj) static int cache_alloc(struct cache *ca) { size_t free; + size_t btree_buckets; struct bucket *b; + int ret = -ENOMEM; + const char *err = NULL; __module_get(THIS_MODULE); kobject_init(&ca->kobj, &bch_cache_ktype); - bio_init(&ca->journal.bio, ca->journal.bio.bi_inline_vecs, 8); + bio_init_inline(&ca->journal.bio, NULL, 8, 0); + /* + * When the cache disk is first registered, ca->sb.njournal_buckets + * is zero, and it is assigned in run_cache_set(). + * + * When ca->sb.njournal_buckets is not zero, journal exists, + * and in bch_journal_replay(), tree node may split. + * The worst situation is all journal buckets are valid journal, + * and all the keys need to replay, so the number of RESERVE_BTREE + * type buckets should be as much as journal buckets. + * + * If the number of RESERVE_BTREE type buckets is too few, the + * bch_allocator_thread() may hang up and unable to allocate + * bucket. The situation is roughly as follows: + * + * 1. In bch_data_insert_keys(), if the operation is not op->replace, + * it will call the bch_journal(), which increments the journal_ref + * counter. This counter is only decremented after bch_btree_insert + * completes. + * + * 2. When calling bch_btree_insert, if the btree needs to split, + * it will call btree_split() and btree_check_reserve() to check + * whether there are enough reserved buckets in the RESERVE_BTREE + * slot. If not enough, bcache_btree_root() will repeatedly retry. + * + * 3. Normally, the bch_allocator_thread is responsible for filling + * the reservation slots from the free_inc bucket list. When the + * free_inc bucket list is exhausted, the bch_allocator_thread + * will call invalidate_buckets() until free_inc is refilled. + * Then bch_allocator_thread calls bch_prio_write() once. and + * bch_prio_write() will call bch_journal_meta() and waits for + * the journal write to complete. + * + * 4. During journal_write, journal_write_unlocked() is be called. + * If journal full occurs, journal_reclaim() and btree_flush_write() + * will be called sequentially, then retry journal_write. + * + * 5. When 2 and 4 occur together, IO will hung up and cannot recover. + * + * Therefore, reserve more RESERVE_BTREE type buckets. + */ + btree_buckets = clamp_t(size_t, ca->sb.nbuckets >> 7, + 32, SB_JOURNAL_BUCKETS); free = roundup_pow_of_two(ca->sb.nbuckets) >> 10; + if (!free) { + ret = -EPERM; + err = "ca->sb.nbuckets is too small"; + goto err_free; + } - if (!init_fifo(&ca->free[RESERVE_BTREE], 8, GFP_KERNEL) || - !init_fifo_exact(&ca->free[RESERVE_PRIO], prio_buckets(ca), GFP_KERNEL) || - !init_fifo(&ca->free[RESERVE_MOVINGGC], free, GFP_KERNEL) || - !init_fifo(&ca->free[RESERVE_NONE], free, GFP_KERNEL) || - !init_fifo(&ca->free_inc, free << 2, GFP_KERNEL) || - !init_heap(&ca->heap, free << 3, GFP_KERNEL) || - !(ca->buckets = vzalloc(sizeof(struct bucket) * - ca->sb.nbuckets)) || - !(ca->prio_buckets = kzalloc(sizeof(uint64_t) * prio_buckets(ca) * - 2, GFP_KERNEL)) || - !(ca->disk_buckets = alloc_bucket_pages(GFP_KERNEL, ca))) - return -ENOMEM; + if (!init_fifo(&ca->free[RESERVE_BTREE], btree_buckets, + GFP_KERNEL)) { + err = "ca->free[RESERVE_BTREE] alloc failed"; + goto err_btree_alloc; + } + + if (!init_fifo_exact(&ca->free[RESERVE_PRIO], prio_buckets(ca), + GFP_KERNEL)) { + err = "ca->free[RESERVE_PRIO] alloc failed"; + goto err_prio_alloc; + } + + if (!init_fifo(&ca->free[RESERVE_MOVINGGC], free, GFP_KERNEL)) { + err = "ca->free[RESERVE_MOVINGGC] alloc failed"; + goto err_movinggc_alloc; + } + + if (!init_fifo(&ca->free[RESERVE_NONE], free, GFP_KERNEL)) { + err = "ca->free[RESERVE_NONE] alloc failed"; + goto err_none_alloc; + } + + if (!init_fifo(&ca->free_inc, free << 2, GFP_KERNEL)) { + err = "ca->free_inc alloc failed"; + goto err_free_inc_alloc; + } + + if (!init_heap(&ca->heap, free << 3, GFP_KERNEL)) { + err = "ca->heap alloc failed"; + goto err_heap_alloc; + } + + ca->buckets = vzalloc(array_size(sizeof(struct bucket), + ca->sb.nbuckets)); + if (!ca->buckets) { + err = "ca->buckets alloc failed"; + goto err_buckets_alloc; + } + + ca->prio_buckets = kzalloc(array3_size(sizeof(uint64_t), + prio_buckets(ca), 2), + GFP_KERNEL); + if (!ca->prio_buckets) { + err = "ca->prio_buckets alloc failed"; + goto err_prio_buckets_alloc; + } + + ca->disk_buckets = alloc_meta_bucket_pages(GFP_KERNEL, &ca->sb); + if (!ca->disk_buckets) { + err = "ca->disk_buckets alloc failed"; + goto err_disk_buckets_alloc; + } ca->prio_last_buckets = ca->prio_buckets + prio_buckets(ca); for_each_bucket(b, ca) atomic_set(&b->pin, 0); - return 0; + +err_disk_buckets_alloc: + kfree(ca->prio_buckets); +err_prio_buckets_alloc: + vfree(ca->buckets); +err_buckets_alloc: + free_heap(&ca->heap); +err_heap_alloc: + free_fifo(&ca->free_inc); +err_free_inc_alloc: + free_fifo(&ca->free[RESERVE_NONE]); +err_none_alloc: + free_fifo(&ca->free[RESERVE_MOVINGGC]); +err_movinggc_alloc: + free_fifo(&ca->free[RESERVE_PRIO]); +err_prio_alloc: + free_fifo(&ca->free[RESERVE_BTREE]); +err_btree_alloc: +err_free: + module_put(THIS_MODULE); + if (err) + pr_notice("error %pg: %s\n", ca->bdev, err); + return ret; } -static int register_cache(struct cache_sb *sb, struct page *sb_page, - struct block_device *bdev, struct cache *ca) +static int register_cache(struct cache_sb *sb, struct cache_sb_disk *sb_disk, + struct file *bdev_file, + struct cache *ca) { - char name[BDEVNAME_SIZE]; const char *err = NULL; /* must be set for any error case */ int ret = 0; memcpy(&ca->sb, sb, sizeof(struct cache_sb)); - ca->bdev = bdev; - ca->bdev->bd_holder = ca; - - bio_init(&ca->sb_bio, ca->sb_bio.bi_inline_vecs, 1); - ca->sb_bio.bi_io_vec[0].bv_page = sb_page; - get_page(sb_page); - - if (blk_queue_discard(bdev_get_queue(ca->bdev))) - ca->discard = CACHE_DISCARD(&ca->sb); + ca->bdev_file = bdev_file; + ca->bdev = file_bdev(bdev_file); + ca->sb_disk = sb_disk; ret = cache_alloc(ca); if (ret != 0) { if (ret == -ENOMEM) err = "cache_alloc(): -ENOMEM"; + else if (ret == -EPERM) + err = "cache_alloc(): cache device is too small"; else err = "cache_alloc(): unknown error"; - goto err; + pr_notice("error %pg: %s\n", file_bdev(bdev_file), err); + /* + * If we failed here, it means ca->kobj is not initialized yet, + * kobject_put() won't be called and there is no chance to + * call fput() to bdev in bch_cache_release(). So + * we explicitly call fput() on the block device here. + */ + fput(bdev_file); + return ret; } - if (kobject_add(&ca->kobj, &part_to_dev(bdev->bd_part)->kobj, "bcache")) { - err = "error calling kobject_add"; + if (kobject_add(&ca->kobj, bdev_kobj(file_bdev(bdev_file)), "bcache")) { + pr_notice("error %pg: error calling kobject_add\n", + file_bdev(bdev_file)); ret = -ENOMEM; goto out; } @@ -1884,134 +2418,326 @@ static int register_cache(struct cache_sb *sb, struct page *sb_page, goto out; } - pr_info("registered cache device %s", bdevname(bdev, name)); + pr_info("registered cache device %pg\n", file_bdev(ca->bdev_file)); out: kobject_put(&ca->kobj); - -err: - if (err) - pr_notice("error opening %s: %s", bdevname(bdev, name), err); - return ret; } /* Global interfaces/init */ -static ssize_t register_bcache(struct kobject *, struct kobj_attribute *, - const char *, size_t); +static ssize_t register_bcache(struct kobject *k, struct kobj_attribute *attr, + const char *buffer, size_t size); +static ssize_t bch_pending_bdevs_cleanup(struct kobject *k, + struct kobj_attribute *attr, + const char *buffer, size_t size); kobj_attribute_write(register, register_bcache); kobj_attribute_write(register_quiet, register_bcache); +kobj_attribute_write(pendings_cleanup, bch_pending_bdevs_cleanup); -static bool bch_is_open_backing(struct block_device *bdev) { +static bool bch_is_open_backing(dev_t dev) +{ struct cache_set *c, *tc; struct cached_dev *dc, *t; list_for_each_entry_safe(c, tc, &bch_cache_sets, list) list_for_each_entry_safe(dc, t, &c->cached_devs, list) - if (dc->bdev == bdev) + if (dc->bdev->bd_dev == dev) return true; list_for_each_entry_safe(dc, t, &uncached_devices, list) - if (dc->bdev == bdev) + if (dc->bdev->bd_dev == dev) return true; return false; } -static bool bch_is_open_cache(struct block_device *bdev) { +static bool bch_is_open_cache(dev_t dev) +{ struct cache_set *c, *tc; - struct cache *ca; - unsigned i; - list_for_each_entry_safe(c, tc, &bch_cache_sets, list) - for_each_cache(ca, c, i) - if (ca->bdev == bdev) - return true; + list_for_each_entry_safe(c, tc, &bch_cache_sets, list) { + struct cache *ca = c->cache; + + if (ca->bdev->bd_dev == dev) + return true; + } + return false; } -static bool bch_is_open(struct block_device *bdev) { - return bch_is_open_cache(bdev) || bch_is_open_backing(bdev); +static bool bch_is_open(dev_t dev) +{ + return bch_is_open_cache(dev) || bch_is_open_backing(dev); +} + +struct async_reg_args { + struct delayed_work reg_work; + char *path; + struct cache_sb *sb; + struct cache_sb_disk *sb_disk; + struct file *bdev_file; + void *holder; +}; + +static void register_bdev_worker(struct work_struct *work) +{ + int fail = false; + struct async_reg_args *args = + container_of(work, struct async_reg_args, reg_work.work); + + mutex_lock(&bch_register_lock); + if (register_bdev(args->sb, args->sb_disk, args->bdev_file, + args->holder) < 0) + fail = true; + mutex_unlock(&bch_register_lock); + + if (fail) + pr_info("error %s: fail to register backing device\n", + args->path); + kfree(args->sb); + kfree(args->path); + kfree(args); + module_put(THIS_MODULE); +} + +static void register_cache_worker(struct work_struct *work) +{ + int fail = false; + struct async_reg_args *args = + container_of(work, struct async_reg_args, reg_work.work); + + /* blkdev_put() will be called in bch_cache_release() */ + if (register_cache(args->sb, args->sb_disk, args->bdev_file, + args->holder)) + fail = true; + + if (fail) + pr_info("error %s: fail to register cache device\n", + args->path); + kfree(args->sb); + kfree(args->path); + kfree(args); + module_put(THIS_MODULE); +} + +static void register_device_async(struct async_reg_args *args) +{ + if (SB_IS_BDEV(args->sb)) + INIT_DELAYED_WORK(&args->reg_work, register_bdev_worker); + else + INIT_DELAYED_WORK(&args->reg_work, register_cache_worker); + + /* 10 jiffies is enough for a delay */ + queue_delayed_work(system_percpu_wq, &args->reg_work, 10); +} + +static void *alloc_holder_object(struct cache_sb *sb) +{ + if (SB_IS_BDEV(sb)) + return kzalloc(sizeof(struct cached_dev), GFP_KERNEL); + return kzalloc(sizeof(struct cache), GFP_KERNEL); } static ssize_t register_bcache(struct kobject *k, struct kobj_attribute *attr, const char *buffer, size_t size) { - ssize_t ret = size; - const char *err = "cannot allocate memory"; + const char *err; char *path = NULL; - struct cache_sb *sb = NULL; - struct block_device *bdev = NULL; - struct page *sb_page = NULL; - + struct cache_sb *sb; + struct cache_sb_disk *sb_disk; + struct file *bdev_file, *bdev_file2; + void *holder = NULL; + ssize_t ret; + bool async_registration = false; + bool quiet = false; + +#ifdef CONFIG_BCACHE_ASYNC_REGISTRATION + async_registration = true; +#endif + + ret = -EBUSY; + err = "failed to reference bcache module"; if (!try_module_get(THIS_MODULE)) - return -EBUSY; + goto out; - if (!(path = kstrndup(buffer, size, GFP_KERNEL)) || - !(sb = kmalloc(sizeof(struct cache_sb), GFP_KERNEL))) - goto err; + /* For latest state of bcache_is_reboot */ + smp_mb(); + err = "bcache is in reboot"; + if (bcache_is_reboot) + goto out_module_put; + + ret = -ENOMEM; + err = "cannot allocate memory"; + path = kstrndup(buffer, size, GFP_KERNEL); + if (!path) + goto out_module_put; + sb = kmalloc(sizeof(struct cache_sb), GFP_KERNEL); + if (!sb) + goto out_free_path; + + ret = -EINVAL; err = "failed to open device"; - bdev = blkdev_get_by_path(strim(path), - FMODE_READ|FMODE_WRITE|FMODE_EXCL, - sb); - if (IS_ERR(bdev)) { - if (bdev == ERR_PTR(-EBUSY)) { - bdev = lookup_bdev(strim(path)); + bdev_file = bdev_file_open_by_path(strim(path), BLK_OPEN_READ, NULL, NULL); + if (IS_ERR(bdev_file)) + goto out_free_sb; + + err = read_super(sb, file_bdev(bdev_file), &sb_disk); + if (err) + goto out_blkdev_put; + + holder = alloc_holder_object(sb); + if (!holder) { + ret = -ENOMEM; + err = "cannot allocate memory"; + goto out_put_sb_folio; + } + + /* Now reopen in exclusive mode with proper holder */ + bdev_file2 = bdev_file_open_by_dev(file_bdev(bdev_file)->bd_dev, + BLK_OPEN_READ | BLK_OPEN_WRITE, holder, NULL); + fput(bdev_file); + bdev_file = bdev_file2; + if (IS_ERR(bdev_file)) { + ret = PTR_ERR(bdev_file); + bdev_file = NULL; + if (ret == -EBUSY) { + dev_t dev; + mutex_lock(&bch_register_lock); - if (!IS_ERR(bdev) && bch_is_open(bdev)) + if (lookup_bdev(strim(path), &dev) == 0 && + bch_is_open(dev)) err = "device already registered"; else err = "device busy"; mutex_unlock(&bch_register_lock); - if (attr == &ksysfs_register_quiet) - goto out; + if (attr == &ksysfs_register_quiet) { + quiet = true; + ret = size; + } } - goto err; + goto out_free_holder; } - err = "failed to set blocksize"; - if (set_blocksize(bdev, 4096)) - goto err_close; + err = "failed to register device"; - err = read_super(sb, bdev, &sb_page); - if (err) - goto err_close; + if (async_registration) { + /* register in asynchronous way */ + struct async_reg_args *args = + kzalloc(sizeof(struct async_reg_args), GFP_KERNEL); - if (SB_IS_BDEV(sb)) { - struct cached_dev *dc = kzalloc(sizeof(*dc), GFP_KERNEL); - if (!dc) - goto err_close; + if (!args) { + ret = -ENOMEM; + err = "cannot allocate memory"; + goto out_free_holder; + } + + args->path = path; + args->sb = sb; + args->sb_disk = sb_disk; + args->bdev_file = bdev_file; + args->holder = holder; + register_device_async(args); + /* No wait and returns to user space */ + goto async_done; + } + if (SB_IS_BDEV(sb)) { mutex_lock(&bch_register_lock); - register_bdev(sb, sb_page, bdev, dc); + ret = register_bdev(sb, sb_disk, bdev_file, holder); mutex_unlock(&bch_register_lock); + /* blkdev_put() will be called in cached_dev_free() */ + if (ret < 0) + goto out_free_sb; } else { - struct cache *ca = kzalloc(sizeof(*ca), GFP_KERNEL); - if (!ca) - goto err_close; - - if (register_cache(sb, sb_page, bdev, ca) != 0) - goto err_close; + /* blkdev_put() will be called in bch_cache_release() */ + ret = register_cache(sb, sb_disk, bdev_file, holder); + if (ret) + goto out_free_sb; } -out: - if (sb_page) - put_page(sb_page); + + kfree(sb); + kfree(path); + module_put(THIS_MODULE); +async_done: + return size; + +out_free_holder: + kfree(holder); +out_put_sb_folio: + folio_put(virt_to_folio(sb_disk)); +out_blkdev_put: + if (bdev_file) + fput(bdev_file); +out_free_sb: kfree(sb); +out_free_path: kfree(path); + path = NULL; +out_module_put: module_put(THIS_MODULE); +out: + if (!quiet) + pr_info("error %s: %s\n", path?path:"", err); return ret; +} -err_close: - blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL); -err: - pr_info("error opening %s: %s", path, err); - ret = -EINVAL; - goto out; + +struct pdev { + struct list_head list; + struct cached_dev *dc; +}; + +static ssize_t bch_pending_bdevs_cleanup(struct kobject *k, + struct kobj_attribute *attr, + const char *buffer, + size_t size) +{ + LIST_HEAD(pending_devs); + ssize_t ret = size; + struct cached_dev *dc, *tdc; + struct pdev *pdev, *tpdev; + struct cache_set *c, *tc; + + mutex_lock(&bch_register_lock); + list_for_each_entry_safe(dc, tdc, &uncached_devices, list) { + pdev = kmalloc(sizeof(struct pdev), GFP_KERNEL); + if (!pdev) + break; + pdev->dc = dc; + list_add(&pdev->list, &pending_devs); + } + + list_for_each_entry_safe(pdev, tpdev, &pending_devs, list) { + char *pdev_set_uuid = pdev->dc->sb.set_uuid; + list_for_each_entry_safe(c, tc, &bch_cache_sets, list) { + char *set_uuid = c->set_uuid; + + if (!memcmp(pdev_set_uuid, set_uuid, 16)) { + list_del(&pdev->list); + kfree(pdev); + break; + } + } + } + mutex_unlock(&bch_register_lock); + + list_for_each_entry_safe(pdev, tpdev, &pending_devs, list) { + pr_info("delete pdev %p\n", pdev); + list_del(&pdev->list); + bcache_device_stop(&pdev->dc->disk); + kfree(pdev); + } + + return ret; } static int bcache_reboot(struct notifier_block *n, unsigned long code, void *x) { + if (bcache_is_reboot) + return NOTIFY_DONE; + if (code == SYS_DOWN || code == SYS_HALT || code == SYS_POWER_OFF) { @@ -2024,22 +2750,57 @@ static int bcache_reboot(struct notifier_block *n, unsigned long code, void *x) mutex_lock(&bch_register_lock); + if (bcache_is_reboot) + goto out; + + /* New registration is rejected since now */ + bcache_is_reboot = true; + /* + * Make registering caller (if there is) on other CPU + * core know bcache_is_reboot set to true earlier + */ + smp_mb(); + if (list_empty(&bch_cache_sets) && list_empty(&uncached_devices)) goto out; - pr_info("Stopping all devices:"); + mutex_unlock(&bch_register_lock); + + pr_info("Stopping all devices:\n"); + /* + * The reason bch_register_lock is not held to call + * bch_cache_set_stop() and bcache_device_stop() is to + * avoid potential deadlock during reboot, because cache + * set or bcache device stopping process will acquire + * bch_register_lock too. + * + * We are safe here because bcache_is_reboot sets to + * true already, register_bcache() will reject new + * registration now. bcache_is_reboot also makes sure + * bcache_reboot() won't be re-entered on by other thread, + * so there is no race in following list iteration by + * list_for_each_entry_safe(). + */ list_for_each_entry_safe(c, tc, &bch_cache_sets, list) bch_cache_set_stop(c); list_for_each_entry_safe(dc, tdc, &uncached_devices, list) bcache_device_stop(&dc->disk); + + /* + * Give an early chance for other kthreads and + * kworkers to stop themselves + */ + schedule(); + /* What's a condition variable? */ while (1) { - long timeout = start + 2 * HZ - jiffies; + long timeout = start + 10 * HZ - jiffies; + mutex_lock(&bch_register_lock); stopped = list_empty(&bch_cache_sets) && list_empty(&uncached_devices); @@ -2051,15 +2812,14 @@ static int bcache_reboot(struct notifier_block *n, unsigned long code, void *x) mutex_unlock(&bch_register_lock); schedule_timeout(timeout); - mutex_lock(&bch_register_lock); } finish_wait(&unregister_wait, &wait); if (stopped) - pr_info("All devices stopped"); + pr_info("All devices stopped\n"); else - pr_notice("Timeout waiting for devices to be closed"); + pr_notice("Timeout waiting for devices to be closed\n"); out: mutex_unlock(&bch_register_lock); } @@ -2080,9 +2840,42 @@ static void bcache_exit(void) kobject_put(bcache_kobj); if (bcache_wq) destroy_workqueue(bcache_wq); + if (bch_journal_wq) + destroy_workqueue(bch_journal_wq); + if (bch_flush_wq) + destroy_workqueue(bch_flush_wq); + bch_btree_exit(); + if (bcache_major) unregister_blkdev(bcache_major, "bcache"); unregister_reboot_notifier(&reboot); + mutex_destroy(&bch_register_lock); +} + +/* Check and fixup module parameters */ +static void check_module_parameters(void) +{ + if (bch_cutoff_writeback_sync == 0) + bch_cutoff_writeback_sync = CUTOFF_WRITEBACK_SYNC; + else if (bch_cutoff_writeback_sync > CUTOFF_WRITEBACK_SYNC_MAX) { + pr_warn("set bch_cutoff_writeback_sync (%u) to max value %u\n", + bch_cutoff_writeback_sync, CUTOFF_WRITEBACK_SYNC_MAX); + bch_cutoff_writeback_sync = CUTOFF_WRITEBACK_SYNC_MAX; + } + + if (bch_cutoff_writeback == 0) + bch_cutoff_writeback = CUTOFF_WRITEBACK; + else if (bch_cutoff_writeback > CUTOFF_WRITEBACK_MAX) { + pr_warn("set bch_cutoff_writeback (%u) to max value %u\n", + bch_cutoff_writeback, CUTOFF_WRITEBACK_MAX); + bch_cutoff_writeback = CUTOFF_WRITEBACK_MAX; + } + + if (bch_cutoff_writeback > bch_cutoff_writeback_sync) { + pr_warn("set bch_cutoff_writeback (%u) to %u\n", + bch_cutoff_writeback, bch_cutoff_writeback_sync); + bch_cutoff_writeback = bch_cutoff_writeback_sync; + } } static int __init bcache_init(void) @@ -2090,32 +2883,78 @@ static int __init bcache_init(void) static const struct attribute *files[] = { &ksysfs_register.attr, &ksysfs_register_quiet.attr, + &ksysfs_pendings_cleanup.attr, NULL }; + check_module_parameters(); + mutex_init(&bch_register_lock); init_waitqueue_head(&unregister_wait); register_reboot_notifier(&reboot); - closure_debug_init(); bcache_major = register_blkdev(0, "bcache"); if (bcache_major < 0) { unregister_reboot_notifier(&reboot); + mutex_destroy(&bch_register_lock); return bcache_major; } - if (!(bcache_wq = alloc_workqueue("bcache", WQ_MEM_RECLAIM, 0)) || - !(bcache_kobj = kobject_create_and_add("bcache", fs_kobj)) || - sysfs_create_files(bcache_kobj, files) || - bch_request_init() || - bch_debug_init(bcache_kobj)) + if (bch_btree_init()) + goto err; + + bcache_wq = alloc_workqueue("bcache", WQ_MEM_RECLAIM | WQ_PERCPU, 0); + if (!bcache_wq) + goto err; + + /* + * Let's not make this `WQ_MEM_RECLAIM` for the following reasons: + * + * 1. It used `system_percpu_wq` before which also does no memory reclaim. + * 2. With `WQ_MEM_RECLAIM` desktop stalls, increased boot times, and + * reduced throughput can be observed. + * + * We still want to user our own queue to not congest the `system_percpu_wq`. + */ + bch_flush_wq = alloc_workqueue("bch_flush", WQ_PERCPU, 0); + if (!bch_flush_wq) + goto err; + + bch_journal_wq = alloc_workqueue("bch_journal", + WQ_MEM_RECLAIM | WQ_PERCPU, 0); + if (!bch_journal_wq) + goto err; + + bcache_kobj = kobject_create_and_add("bcache", fs_kobj); + if (!bcache_kobj) + goto err; + + if (bch_request_init() || + sysfs_create_files(bcache_kobj, files)) goto err; + bch_debug_init(); + + bcache_is_reboot = false; + return 0; err: bcache_exit(); return -ENOMEM; } +/* + * Module hooks + */ module_exit(bcache_exit); module_init(bcache_init); + +module_param(bch_cutoff_writeback, uint, 0); +MODULE_PARM_DESC(bch_cutoff_writeback, "threshold to cutoff writeback"); + +module_param(bch_cutoff_writeback_sync, uint, 0); +MODULE_PARM_DESC(bch_cutoff_writeback_sync, "hard threshold to cutoff writeback"); + +MODULE_DESCRIPTION("Bcache: a Linux block layer cache"); +MODULE_AUTHOR("Kent Overstreet <kent.overstreet@gmail.com>"); +MODULE_LICENSE("GPL"); |