diff options
Diffstat (limited to 'fs')
43 files changed, 852 insertions, 653 deletions
diff --git a/fs/bcachefs/alloc_foreground.c b/fs/bcachefs/alloc_foreground.c index 7c930ef77380..effafc3e0ced 100644 --- a/fs/bcachefs/alloc_foreground.c +++ b/fs/bcachefs/alloc_foreground.c @@ -1425,6 +1425,8 @@ alloc_done: open_bucket_for_each(c, &wp->ptrs, ob, i) wp->sectors_free = min(wp->sectors_free, ob->sectors_free); + wp->sectors_free = rounddown(wp->sectors_free, block_sectors(c)); + BUG_ON(!wp->sectors_free || wp->sectors_free == UINT_MAX); return 0; diff --git a/fs/bcachefs/alloc_foreground.h b/fs/bcachefs/alloc_foreground.h index 69ec6a012898..4c1e33cf57c0 100644 --- a/fs/bcachefs/alloc_foreground.h +++ b/fs/bcachefs/alloc_foreground.h @@ -110,7 +110,9 @@ static inline void bch2_alloc_sectors_done_inlined(struct bch_fs *c, struct writ unsigned i; open_bucket_for_each(c, &wp->ptrs, ob, i) - ob_push(c, !ob->sectors_free ? &ptrs : &keep, ob); + ob_push(c, ob->sectors_free < block_sectors(c) + ? &ptrs + : &keep, ob); wp->ptrs = keep; mutex_unlock(&wp->lock); diff --git a/fs/bcachefs/bcachefs_format.h b/fs/bcachefs/bcachefs_format.h index a3db328dee31..d6e4a496f02b 100644 --- a/fs/bcachefs/bcachefs_format.h +++ b/fs/bcachefs/bcachefs_format.h @@ -366,6 +366,10 @@ static inline void bkey_init(struct bkey *k) #define __BKEY_PADDED(key, pad) \ struct bkey_i key; __u64 key ## _pad[pad] +enum bch_bkey_type_flags { + BKEY_TYPE_strict_btree_checks = BIT(0), +}; + /* * - DELETED keys are used internally to mark keys that should be ignored but * override keys in composition order. Their version number is ignored. @@ -383,46 +387,46 @@ static inline void bkey_init(struct bkey *k) * * - WHITEOUT: for hash table btrees */ -#define BCH_BKEY_TYPES() \ - x(deleted, 0) \ - x(whiteout, 1) \ - x(error, 2) \ - x(cookie, 3) \ - x(hash_whiteout, 4) \ - x(btree_ptr, 5) \ - x(extent, 6) \ - x(reservation, 7) \ - x(inode, 8) \ - x(inode_generation, 9) \ - x(dirent, 10) \ - x(xattr, 11) \ - x(alloc, 12) \ - x(quota, 13) \ - x(stripe, 14) \ - x(reflink_p, 15) \ - x(reflink_v, 16) \ - x(inline_data, 17) \ - x(btree_ptr_v2, 18) \ - x(indirect_inline_data, 19) \ - x(alloc_v2, 20) \ - x(subvolume, 21) \ - x(snapshot, 22) \ - x(inode_v2, 23) \ - x(alloc_v3, 24) \ - x(set, 25) \ - x(lru, 26) \ - x(alloc_v4, 27) \ - x(backpointer, 28) \ - x(inode_v3, 29) \ - x(bucket_gens, 30) \ - x(snapshot_tree, 31) \ - x(logged_op_truncate, 32) \ - x(logged_op_finsert, 33) \ - x(accounting, 34) \ - x(inode_alloc_cursor, 35) +#define BCH_BKEY_TYPES() \ + x(deleted, 0, 0) \ + x(whiteout, 1, 0) \ + x(error, 2, 0) \ + x(cookie, 3, 0) \ + x(hash_whiteout, 4, BKEY_TYPE_strict_btree_checks) \ + x(btree_ptr, 5, BKEY_TYPE_strict_btree_checks) \ + x(extent, 6, BKEY_TYPE_strict_btree_checks) \ + x(reservation, 7, BKEY_TYPE_strict_btree_checks) \ + x(inode, 8, BKEY_TYPE_strict_btree_checks) \ + x(inode_generation, 9, BKEY_TYPE_strict_btree_checks) \ + x(dirent, 10, BKEY_TYPE_strict_btree_checks) \ + x(xattr, 11, BKEY_TYPE_strict_btree_checks) \ + x(alloc, 12, BKEY_TYPE_strict_btree_checks) \ + x(quota, 13, BKEY_TYPE_strict_btree_checks) \ + x(stripe, 14, BKEY_TYPE_strict_btree_checks) \ + x(reflink_p, 15, BKEY_TYPE_strict_btree_checks) \ + x(reflink_v, 16, BKEY_TYPE_strict_btree_checks) \ + x(inline_data, 17, BKEY_TYPE_strict_btree_checks) \ + x(btree_ptr_v2, 18, BKEY_TYPE_strict_btree_checks) \ + x(indirect_inline_data, 19, BKEY_TYPE_strict_btree_checks) \ + x(alloc_v2, 20, BKEY_TYPE_strict_btree_checks) \ + x(subvolume, 21, BKEY_TYPE_strict_btree_checks) \ + x(snapshot, 22, BKEY_TYPE_strict_btree_checks) \ + x(inode_v2, 23, BKEY_TYPE_strict_btree_checks) \ + x(alloc_v3, 24, BKEY_TYPE_strict_btree_checks) \ + x(set, 25, 0) \ + x(lru, 26, BKEY_TYPE_strict_btree_checks) \ + x(alloc_v4, 27, BKEY_TYPE_strict_btree_checks) \ + x(backpointer, 28, BKEY_TYPE_strict_btree_checks) \ + x(inode_v3, 29, BKEY_TYPE_strict_btree_checks) \ + x(bucket_gens, 30, BKEY_TYPE_strict_btree_checks) \ + x(snapshot_tree, 31, BKEY_TYPE_strict_btree_checks) \ + x(logged_op_truncate, 32, BKEY_TYPE_strict_btree_checks) \ + x(logged_op_finsert, 33, BKEY_TYPE_strict_btree_checks) \ + x(accounting, 34, BKEY_TYPE_strict_btree_checks) \ + x(inode_alloc_cursor, 35, BKEY_TYPE_strict_btree_checks) enum bch_bkey_type { -#define x(name, nr) KEY_TYPE_##name = nr, +#define x(name, nr, ...) KEY_TYPE_##name = nr, BCH_BKEY_TYPES() #undef x KEY_TYPE_MAX, @@ -863,6 +867,7 @@ LE64_BITMASK(BCH_SB_VERSION_INCOMPAT_ALLOWED, LE64_BITMASK(BCH_SB_SHARD_INUMS_NBITS, struct bch_sb, flags[6], 0, 4); LE64_BITMASK(BCH_SB_WRITE_ERROR_TIMEOUT,struct bch_sb, flags[6], 4, 14); LE64_BITMASK(BCH_SB_CSUM_ERR_RETRY_NR, struct bch_sb, flags[6], 14, 20); +LE64_BITMASK(BCH_SB_CASEFOLD, struct bch_sb, flags[6], 22, 23); static inline __u64 BCH_SB_COMPRESSION_TYPE(const struct bch_sb *sb) { diff --git a/fs/bcachefs/bkey_methods.c b/fs/bcachefs/bkey_methods.c index 15c93576b5c2..00d05ccfaf73 100644 --- a/fs/bcachefs/bkey_methods.c +++ b/fs/bcachefs/bkey_methods.c @@ -21,7 +21,7 @@ #include "xattr.h" const char * const bch2_bkey_types[] = { -#define x(name, nr) #name, +#define x(name, nr, ...) #name, BCH_BKEY_TYPES() #undef x NULL @@ -115,7 +115,7 @@ static bool key_type_set_merge(struct bch_fs *c, struct bkey_s l, struct bkey_s_ }) const struct bkey_ops bch2_bkey_ops[] = { -#define x(name, nr) [KEY_TYPE_##name] = bch2_bkey_ops_##name, +#define x(name, nr, ...) [KEY_TYPE_##name] = bch2_bkey_ops_##name, BCH_BKEY_TYPES() #undef x }; @@ -155,6 +155,12 @@ static u64 bch2_key_types_allowed[] = { #undef x }; +static const enum bch_bkey_type_flags bch2_bkey_type_flags[] = { +#define x(name, nr, flags) [KEY_TYPE_##name] = flags, + BCH_BKEY_TYPES() +#undef x +}; + const char *bch2_btree_node_type_str(enum btree_node_type type) { return type == BKEY_TYPE_btree ? "internal btree node" : bch2_btree_id_str(type - 1); @@ -177,8 +183,18 @@ int __bch2_bkey_validate(struct bch_fs *c, struct bkey_s_c k, if (type >= BKEY_TYPE_NR) return 0; - bkey_fsck_err_on(k.k->type < KEY_TYPE_MAX && - (type == BKEY_TYPE_btree || (from.flags & BCH_VALIDATE_commit)) && + enum bch_bkey_type_flags bkey_flags = k.k->type < KEY_TYPE_MAX + ? bch2_bkey_type_flags[k.k->type] + : 0; + + bool strict_key_type_allowed = + (from.flags & BCH_VALIDATE_commit) || + type == BKEY_TYPE_btree || + (from.btree < BTREE_ID_NR && + (bkey_flags & BKEY_TYPE_strict_btree_checks)); + + bkey_fsck_err_on(strict_key_type_allowed && + k.k->type < KEY_TYPE_MAX && !(bch2_key_types_allowed[type] & BIT_ULL(k.k->type)), c, bkey_invalid_type_for_btree, "invalid key type for btree %s (%s)", diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c index e34e9598ef25..59fa527ac685 100644 --- a/fs/bcachefs/btree_iter.c +++ b/fs/bcachefs/btree_iter.c @@ -2577,7 +2577,10 @@ struct bkey_s_c bch2_btree_iter_peek_prev_min(struct btree_trans *trans, struct struct bpos end) { if ((iter->flags & (BTREE_ITER_is_extents|BTREE_ITER_filter_snapshots)) && - !bkey_eq(iter->pos, POS_MAX)) { + !bkey_eq(iter->pos, POS_MAX) && + !((iter->flags & BTREE_ITER_is_extents) && + iter->pos.offset == U64_MAX)) { + /* * bkey_start_pos(), for extents, is not monotonically * increasing until after filtering for snapshots: @@ -2602,7 +2605,7 @@ struct bkey_s_c bch2_btree_iter_peek_prev_min(struct btree_trans *trans, struct bch2_trans_verify_not_unlocked_or_in_restart(trans); bch2_btree_iter_verify_entry_exit(iter); - EBUG_ON((iter->flags & BTREE_ITER_filter_snapshots) && bpos_eq(end, POS_MIN)); + EBUG_ON((iter->flags & BTREE_ITER_filter_snapshots) && iter->pos.inode != end.inode); int ret = trans_maybe_inject_restart(trans, _RET_IP_); if (unlikely(ret)) { diff --git a/fs/bcachefs/dirent.c b/fs/bcachefs/dirent.c index 8488a7578115..92ee59d9e00e 100644 --- a/fs/bcachefs/dirent.c +++ b/fs/bcachefs/dirent.c @@ -13,8 +13,8 @@ #include <linux/dcache.h> -static int bch2_casefold(struct btree_trans *trans, const struct bch_hash_info *info, - const struct qstr *str, struct qstr *out_cf) +int bch2_casefold(struct btree_trans *trans, const struct bch_hash_info *info, + const struct qstr *str, struct qstr *out_cf) { *out_cf = (struct qstr) QSTR_INIT(NULL, 0); @@ -35,18 +35,6 @@ static int bch2_casefold(struct btree_trans *trans, const struct bch_hash_info * #endif } -static inline int bch2_maybe_casefold(struct btree_trans *trans, - const struct bch_hash_info *info, - const struct qstr *str, struct qstr *out_cf) -{ - if (likely(!info->cf_encoding)) { - *out_cf = *str; - return 0; - } else { - return bch2_casefold(trans, info, str, out_cf); - } -} - static unsigned bch2_dirent_name_bytes(struct bkey_s_c_dirent d) { if (bkey_val_bytes(d.k) < offsetof(struct bch_dirent, d_name)) diff --git a/fs/bcachefs/dirent.h b/fs/bcachefs/dirent.h index 0880772b80a9..9838a7ba7ed1 100644 --- a/fs/bcachefs/dirent.h +++ b/fs/bcachefs/dirent.h @@ -23,6 +23,21 @@ struct bch_fs; struct bch_hash_info; struct bch_inode_info; +int bch2_casefold(struct btree_trans *, const struct bch_hash_info *, + const struct qstr *, struct qstr *); + +static inline int bch2_maybe_casefold(struct btree_trans *trans, + const struct bch_hash_info *info, + const struct qstr *str, struct qstr *out_cf) +{ + if (likely(!info->cf_encoding)) { + *out_cf = *str; + return 0; + } else { + return bch2_casefold(trans, info, str, out_cf); + } +} + struct qstr bch2_dirent_get_name(struct bkey_s_c_dirent d); static inline unsigned dirent_val_u64s(unsigned len, unsigned cf_len) diff --git a/fs/bcachefs/error.c b/fs/bcachefs/error.c index baf5dfb32298..925b0b54ea2f 100644 --- a/fs/bcachefs/error.c +++ b/fs/bcachefs/error.c @@ -272,9 +272,6 @@ static struct fsck_err_state *fsck_err_get(struct bch_fs *c, { struct fsck_err_state *s; - if (!test_bit(BCH_FS_fsck_running, &c->flags)) - return NULL; - list_for_each_entry(s, &c->fsck_error_msgs, list) if (s->id == id) { /* @@ -639,14 +636,14 @@ int __bch2_bkey_fsck_err(struct bch_fs *c, return ret; } -void bch2_flush_fsck_errs(struct bch_fs *c) +static void __bch2_flush_fsck_errs(struct bch_fs *c, bool print) { struct fsck_err_state *s, *n; mutex_lock(&c->fsck_error_msgs_lock); list_for_each_entry_safe(s, n, &c->fsck_error_msgs, list) { - if (s->ratelimited && s->last_msg) + if (print && s->ratelimited && s->last_msg) bch_err(c, "Saw %llu errors like:\n %s", s->nr, s->last_msg); list_del(&s->list); @@ -657,6 +654,16 @@ void bch2_flush_fsck_errs(struct bch_fs *c) mutex_unlock(&c->fsck_error_msgs_lock); } +void bch2_flush_fsck_errs(struct bch_fs *c) +{ + __bch2_flush_fsck_errs(c, true); +} + +void bch2_free_fsck_errs(struct bch_fs *c) +{ + __bch2_flush_fsck_errs(c, false); +} + int bch2_inum_offset_err_msg_trans(struct btree_trans *trans, struct printbuf *out, subvol_inum inum, u64 offset) { diff --git a/fs/bcachefs/error.h b/fs/bcachefs/error.h index d0d024dc714b..4a364fd44abe 100644 --- a/fs/bcachefs/error.h +++ b/fs/bcachefs/error.h @@ -93,6 +93,7 @@ int __bch2_fsck_err(struct bch_fs *, struct btree_trans *, _flags, BCH_FSCK_ERR_##_err_type, __VA_ARGS__) void bch2_flush_fsck_errs(struct bch_fs *); +void bch2_free_fsck_errs(struct bch_fs *); #define fsck_err_wrap(_do) \ ({ \ diff --git a/fs/bcachefs/fs-ioctl.c b/fs/bcachefs/fs-ioctl.c index 14886e1d4d6d..a82dfce9e4ad 100644 --- a/fs/bcachefs/fs-ioctl.c +++ b/fs/bcachefs/fs-ioctl.c @@ -21,206 +21,6 @@ #define FSOP_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ #define FSOP_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ -struct flags_set { - unsigned mask; - unsigned flags; - - unsigned projid; - - bool set_projinherit; - bool projinherit; -}; - -static int bch2_inode_flags_set(struct btree_trans *trans, - struct bch_inode_info *inode, - struct bch_inode_unpacked *bi, - void *p) -{ - struct bch_fs *c = inode->v.i_sb->s_fs_info; - /* - * We're relying on btree locking here for exclusion with other ioctl - * calls - use the flags in the btree (@bi), not inode->i_flags: - */ - struct flags_set *s = p; - unsigned newflags = s->flags; - unsigned oldflags = bi->bi_flags & s->mask; - - if (((newflags ^ oldflags) & (BCH_INODE_append|BCH_INODE_immutable)) && - !capable(CAP_LINUX_IMMUTABLE)) - return -EPERM; - - if (!S_ISREG(bi->bi_mode) && - !S_ISDIR(bi->bi_mode) && - (newflags & (BCH_INODE_nodump|BCH_INODE_noatime)) != newflags) - return -EINVAL; - - if ((newflags ^ oldflags) & BCH_INODE_casefolded) { -#ifdef CONFIG_UNICODE - int ret = 0; - /* Not supported on individual files. */ - if (!S_ISDIR(bi->bi_mode)) - return -EOPNOTSUPP; - - /* - * Make sure the dir is empty, as otherwise we'd need to - * rehash everything and update the dirent keys. - */ - ret = bch2_empty_dir_trans(trans, inode_inum(inode)); - if (ret < 0) - return ret; - - ret = bch2_request_incompat_feature(c, bcachefs_metadata_version_casefolding); - if (ret) - return ret; - - bch2_check_set_feature(c, BCH_FEATURE_casefolding); -#else - printk(KERN_ERR "Cannot use casefolding on a kernel without CONFIG_UNICODE\n"); - return -EOPNOTSUPP; -#endif - } - - if (s->set_projinherit) { - bi->bi_fields_set &= ~(1 << Inode_opt_project); - bi->bi_fields_set |= ((int) s->projinherit << Inode_opt_project); - } - - bi->bi_flags &= ~s->mask; - bi->bi_flags |= newflags; - - bi->bi_ctime = timespec_to_bch2_time(c, current_time(&inode->v)); - return 0; -} - -static int bch2_ioc_getflags(struct bch_inode_info *inode, int __user *arg) -{ - unsigned flags = map_flags(bch_flags_to_uflags, inode->ei_inode.bi_flags); - - return put_user(flags, arg); -} - -static int bch2_ioc_setflags(struct bch_fs *c, - struct file *file, - struct bch_inode_info *inode, - void __user *arg) -{ - struct flags_set s = { .mask = map_defined(bch_flags_to_uflags) }; - unsigned uflags; - int ret; - - if (get_user(uflags, (int __user *) arg)) - return -EFAULT; - - s.flags = map_flags_rev(bch_flags_to_uflags, uflags); - if (uflags) - return -EOPNOTSUPP; - - ret = mnt_want_write_file(file); - if (ret) - return ret; - - inode_lock(&inode->v); - if (!inode_owner_or_capable(file_mnt_idmap(file), &inode->v)) { - ret = -EACCES; - goto setflags_out; - } - - mutex_lock(&inode->ei_update_lock); - ret = bch2_subvol_is_ro(c, inode->ei_inum.subvol) ?: - bch2_write_inode(c, inode, bch2_inode_flags_set, &s, - ATTR_CTIME); - mutex_unlock(&inode->ei_update_lock); - -setflags_out: - inode_unlock(&inode->v); - mnt_drop_write_file(file); - return ret; -} - -static int bch2_ioc_fsgetxattr(struct bch_inode_info *inode, - struct fsxattr __user *arg) -{ - struct fsxattr fa = { 0 }; - - fa.fsx_xflags = map_flags(bch_flags_to_xflags, inode->ei_inode.bi_flags); - - if (inode->ei_inode.bi_fields_set & (1 << Inode_opt_project)) - fa.fsx_xflags |= FS_XFLAG_PROJINHERIT; - - fa.fsx_projid = inode->ei_qid.q[QTYP_PRJ]; - - if (copy_to_user(arg, &fa, sizeof(fa))) - return -EFAULT; - - return 0; -} - -static int fssetxattr_inode_update_fn(struct btree_trans *trans, - struct bch_inode_info *inode, - struct bch_inode_unpacked *bi, - void *p) -{ - struct flags_set *s = p; - - if (s->projid != bi->bi_project) { - bi->bi_fields_set |= 1U << Inode_opt_project; - bi->bi_project = s->projid; - } - - return bch2_inode_flags_set(trans, inode, bi, p); -} - -static int bch2_ioc_fssetxattr(struct bch_fs *c, - struct file *file, - struct bch_inode_info *inode, - struct fsxattr __user *arg) -{ - struct flags_set s = { .mask = map_defined(bch_flags_to_xflags) }; - struct fsxattr fa; - int ret; - - if (copy_from_user(&fa, arg, sizeof(fa))) - return -EFAULT; - - s.set_projinherit = true; - s.projinherit = (fa.fsx_xflags & FS_XFLAG_PROJINHERIT) != 0; - fa.fsx_xflags &= ~FS_XFLAG_PROJINHERIT; - - s.flags = map_flags_rev(bch_flags_to_xflags, fa.fsx_xflags); - if (fa.fsx_xflags) - return -EOPNOTSUPP; - - if (fa.fsx_projid >= U32_MAX) - return -EINVAL; - - /* - * inode fields accessible via the xattr interface are stored with a +1 - * bias, so that 0 means unset: - */ - s.projid = fa.fsx_projid + 1; - - ret = mnt_want_write_file(file); - if (ret) - return ret; - - inode_lock(&inode->v); - if (!inode_owner_or_capable(file_mnt_idmap(file), &inode->v)) { - ret = -EACCES; - goto err; - } - - mutex_lock(&inode->ei_update_lock); - ret = bch2_subvol_is_ro(c, inode->ei_inum.subvol) ?: - bch2_set_projid(c, inode, fa.fsx_projid) ?: - bch2_write_inode(c, inode, fssetxattr_inode_update_fn, &s, - ATTR_CTIME); - mutex_unlock(&inode->ei_update_lock); -err: - inode_unlock(&inode->v); - mnt_drop_write_file(file); - return ret; -} - static int bch2_reinherit_attrs_fn(struct btree_trans *trans, struct bch_inode_info *inode, struct bch_inode_unpacked *bi, @@ -558,23 +358,6 @@ long bch2_fs_file_ioctl(struct file *file, unsigned cmd, unsigned long arg) long ret; switch (cmd) { - case FS_IOC_GETFLAGS: - ret = bch2_ioc_getflags(inode, (int __user *) arg); - break; - - case FS_IOC_SETFLAGS: - ret = bch2_ioc_setflags(c, file, inode, (int __user *) arg); - break; - - case FS_IOC_FSGETXATTR: - ret = bch2_ioc_fsgetxattr(inode, (void __user *) arg); - break; - - case FS_IOC_FSSETXATTR: - ret = bch2_ioc_fssetxattr(c, file, inode, - (void __user *) arg); - break; - case BCHFS_IOC_REINHERIT_ATTRS: ret = bch2_ioc_reinherit_attrs(c, file, inode, (void __user *) arg); diff --git a/fs/bcachefs/fs-ioctl.h b/fs/bcachefs/fs-ioctl.h index ecd3bfdcde21..a657e4994b71 100644 --- a/fs/bcachefs/fs-ioctl.h +++ b/fs/bcachefs/fs-ioctl.h @@ -2,81 +2,6 @@ #ifndef _BCACHEFS_FS_IOCTL_H #define _BCACHEFS_FS_IOCTL_H -/* Inode flags: */ - -/* bcachefs inode flags -> vfs inode flags: */ -static const __maybe_unused unsigned bch_flags_to_vfs[] = { - [__BCH_INODE_sync] = S_SYNC, - [__BCH_INODE_immutable] = S_IMMUTABLE, - [__BCH_INODE_append] = S_APPEND, - [__BCH_INODE_noatime] = S_NOATIME, - [__BCH_INODE_casefolded] = S_CASEFOLD, -}; - -/* bcachefs inode flags -> FS_IOC_GETFLAGS: */ -static const __maybe_unused unsigned bch_flags_to_uflags[] = { - [__BCH_INODE_sync] = FS_SYNC_FL, - [__BCH_INODE_immutable] = FS_IMMUTABLE_FL, - [__BCH_INODE_append] = FS_APPEND_FL, - [__BCH_INODE_nodump] = FS_NODUMP_FL, - [__BCH_INODE_noatime] = FS_NOATIME_FL, - [__BCH_INODE_casefolded] = FS_CASEFOLD_FL, -}; - -/* bcachefs inode flags -> FS_IOC_FSGETXATTR: */ -static const __maybe_unused unsigned bch_flags_to_xflags[] = { - [__BCH_INODE_sync] = FS_XFLAG_SYNC, - [__BCH_INODE_immutable] = FS_XFLAG_IMMUTABLE, - [__BCH_INODE_append] = FS_XFLAG_APPEND, - [__BCH_INODE_nodump] = FS_XFLAG_NODUMP, - [__BCH_INODE_noatime] = FS_XFLAG_NOATIME, - //[__BCH_INODE_PROJINHERIT] = FS_XFLAG_PROJINHERIT; -}; - -#define set_flags(_map, _in, _out) \ -do { \ - unsigned _i; \ - \ - for (_i = 0; _i < ARRAY_SIZE(_map); _i++) \ - if ((_in) & (1 << _i)) \ - (_out) |= _map[_i]; \ - else \ - (_out) &= ~_map[_i]; \ -} while (0) - -#define map_flags(_map, _in) \ -({ \ - unsigned _out = 0; \ - \ - set_flags(_map, _in, _out); \ - _out; \ -}) - -#define map_flags_rev(_map, _in) \ -({ \ - unsigned _i, _out = 0; \ - \ - for (_i = 0; _i < ARRAY_SIZE(_map); _i++) \ - if ((_in) & _map[_i]) { \ - (_out) |= 1 << _i; \ - (_in) &= ~_map[_i]; \ - } \ - (_out); \ -}) - -#define map_defined(_map) \ -({ \ - unsigned _in = ~0; \ - \ - map_flags_rev(_map, _in); \ -}) - -/* Set VFS inode flags from bcachefs inode: */ -static inline void bch2_inode_flags_to_vfs(struct bch_inode_info *inode) -{ - set_flags(bch_flags_to_vfs, inode->ei_inode.bi_flags, inode->v.i_flags); -} - long bch2_fs_file_ioctl(struct file *, unsigned, unsigned long); long bch2_compat_fs_ioctl(struct file *, unsigned, unsigned long); diff --git a/fs/bcachefs/fs.c b/fs/bcachefs/fs.c index 5a41b1a8e54f..0f1d61aab90b 100644 --- a/fs/bcachefs/fs.c +++ b/fs/bcachefs/fs.c @@ -33,6 +33,7 @@ #include <linux/backing-dev.h> #include <linux/exportfs.h> #include <linux/fiemap.h> +#include <linux/fileattr.h> #include <linux/fs_context.h> #include <linux/module.h> #include <linux/pagemap.h> @@ -51,6 +52,22 @@ static void bch2_vfs_inode_init(struct btree_trans *, subvol_inum, struct bch_inode_unpacked *, struct bch_subvolume *); +/* Set VFS inode flags from bcachefs inode: */ +static inline void bch2_inode_flags_to_vfs(struct bch_fs *c, struct bch_inode_info *inode) +{ + static const __maybe_unused unsigned bch_flags_to_vfs[] = { + [__BCH_INODE_sync] = S_SYNC, + [__BCH_INODE_immutable] = S_IMMUTABLE, + [__BCH_INODE_append] = S_APPEND, + [__BCH_INODE_noatime] = S_NOATIME, + }; + + set_flags(bch_flags_to_vfs, inode->ei_inode.bi_flags, inode->v.i_flags); + + if (bch2_inode_casefold(c, &inode->ei_inode)) + inode->v.i_flags |= S_CASEFOLD; +} + void bch2_inode_update_after_write(struct btree_trans *trans, struct bch_inode_info *inode, struct bch_inode_unpacked *bi, @@ -79,7 +96,7 @@ void bch2_inode_update_after_write(struct btree_trans *trans, inode->ei_inode = *bi; - bch2_inode_flags_to_vfs(inode); + bch2_inode_flags_to_vfs(c, inode); } int __must_check bch2_write_inode(struct bch_fs *c, @@ -631,13 +648,18 @@ static struct bch_inode_info *bch2_lookup_trans(struct btree_trans *trans, const struct qstr *name) { struct bch_fs *c = trans->c; - struct btree_iter dirent_iter = {}; subvol_inum inum = {}; struct printbuf buf = PRINTBUF; + struct qstr lookup_name; + int ret = bch2_maybe_casefold(trans, dir_hash_info, name, &lookup_name); + if (ret) + return ERR_PTR(ret); + + struct btree_iter dirent_iter = {}; struct bkey_s_c k = bch2_hash_lookup(trans, &dirent_iter, bch2_dirent_hash_desc, - dir_hash_info, dir, name, 0); - int ret = bkey_err(k); + dir_hash_info, dir, &lookup_name, 0); + ret = bkey_err(k); if (ret) return ERR_PTR(ret); @@ -825,6 +847,11 @@ int __bch2_unlink(struct inode *vdir, struct dentry *dentry, */ set_nlink(&inode->v, 0); } + + if (IS_CASEFOLDED(vdir)) { + d_invalidate(dentry); + d_prune_aliases(&inode->v); + } err: bch2_trans_put(trans); bch2_unlock_inodes(INODE_UPDATE_LOCK, dir, inode); @@ -1235,10 +1262,20 @@ static int bch2_tmpfile(struct mnt_idmap *idmap, return finish_open_simple(file, 0); } +struct bch_fiemap_extent { + struct bkey_buf kbuf; + unsigned flags; +}; + static int bch2_fill_extent(struct bch_fs *c, struct fiemap_extent_info *info, - struct bkey_s_c k, unsigned flags) + struct bch_fiemap_extent *fe) { + struct bkey_s_c k = bkey_i_to_s_c(fe->kbuf.k); + unsigned flags = fe->flags; + + BUG_ON(!k.k->size); + if (bkey_extent_is_direct_data(k.k)) { struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k); const union bch_extent_entry *entry; @@ -1291,110 +1328,223 @@ static int bch2_fill_extent(struct bch_fs *c, } } -static int bch2_fiemap(struct inode *vinode, struct fiemap_extent_info *info, - u64 start, u64 len) +/* + * Scan a range of an inode for data in pagecache. + * + * Intended to be retryable, so don't modify the output params until success is + * imminent. + */ +static int +bch2_fiemap_hole_pagecache(struct inode *vinode, u64 *start, u64 *end, + bool nonblock) { - struct bch_fs *c = vinode->i_sb->s_fs_info; - struct bch_inode_info *ei = to_bch_ei(vinode); - struct btree_trans *trans; - struct btree_iter iter; - struct bkey_s_c k; - struct bkey_buf cur, prev; - bool have_extent = false; - int ret = 0; + loff_t dstart, dend; - ret = fiemap_prep(&ei->v, info, start, &len, FIEMAP_FLAG_SYNC); - if (ret) + dstart = bch2_seek_pagecache_data(vinode, *start, *end, 0, nonblock); + if (dstart < 0) + return dstart; + + if (dstart == *end) { + *start = dstart; + return 0; + } + + dend = bch2_seek_pagecache_hole(vinode, dstart, *end, 0, nonblock); + if (dend < 0) + return dend; + + /* race */ + BUG_ON(dstart == dend); + + *start = dstart; + *end = dend; + return 0; +} + +/* + * Scan a range of pagecache that corresponds to a file mapping hole in the + * extent btree. If data is found, fake up an extent key so it looks like a + * delalloc extent to the rest of the fiemap processing code. + */ +static int +bch2_next_fiemap_pagecache_extent(struct btree_trans *trans, struct bch_inode_info *inode, + u64 start, u64 end, struct bch_fiemap_extent *cur) +{ + struct bch_fs *c = trans->c; + struct bkey_i_extent *delextent; + struct bch_extent_ptr ptr = {}; + loff_t dstart = start << 9, dend = end << 9; + int ret; + + /* + * We hold btree locks here so we cannot block on folio locks without + * dropping trans locks first. Run a nonblocking scan for the common + * case of no folios over holes and fall back on failure. + * + * Note that dropping locks like this is technically racy against + * writeback inserting to the extent tree, but a non-sync fiemap scan is + * fundamentally racy with writeback anyways. Therefore, just report the + * range as delalloc regardless of whether we have to cycle trans locks. + */ + ret = bch2_fiemap_hole_pagecache(&inode->v, &dstart, &dend, true); + if (ret == -EAGAIN) + ret = drop_locks_do(trans, + bch2_fiemap_hole_pagecache(&inode->v, &dstart, &dend, false)); + if (ret < 0) return ret; - struct bpos end = POS(ei->v.i_ino, (start + len) >> 9); - if (start + len < start) - return -EINVAL; + /* + * Create a fake extent key in the buffer. We have to add a dummy extent + * pointer for the fill code to add an extent entry. It's explicitly + * zeroed to reflect delayed allocation (i.e. phys offset 0). + */ + bch2_bkey_buf_realloc(&cur->kbuf, c, sizeof(*delextent) / sizeof(u64)); + delextent = bkey_extent_init(cur->kbuf.k); + delextent->k.p = POS(inode->ei_inum.inum, dend >> 9); + delextent->k.size = (dend - dstart) >> 9; + bch2_bkey_append_ptr(&delextent->k_i, ptr); - start >>= 9; + cur->flags = FIEMAP_EXTENT_DELALLOC; - bch2_bkey_buf_init(&cur); - bch2_bkey_buf_init(&prev); - trans = bch2_trans_get(c); + return 0; +} + +static int bch2_next_fiemap_extent(struct btree_trans *trans, + struct bch_inode_info *inode, + u64 start, u64 end, + struct bch_fiemap_extent *cur) +{ + u32 snapshot; + int ret = bch2_subvolume_get_snapshot(trans, inode->ei_inum.subvol, &snapshot); + if (ret) + return ret; + struct btree_iter iter; bch2_trans_iter_init(trans, &iter, BTREE_ID_extents, - POS(ei->v.i_ino, start), 0); + SPOS(inode->ei_inum.inum, start, snapshot), 0); - while (!ret || bch2_err_matches(ret, BCH_ERR_transaction_restart)) { - enum btree_id data_btree = BTREE_ID_extents; + struct bkey_s_c k = + bch2_btree_iter_peek_max(trans, &iter, POS(inode->ei_inum.inum, end)); + ret = bkey_err(k); + if (ret) + goto err; - bch2_trans_begin(trans); + ret = bch2_next_fiemap_pagecache_extent(trans, inode, start, end, cur); + if (ret) + goto err; - u32 snapshot; - ret = bch2_subvolume_get_snapshot(trans, ei->ei_inum.subvol, &snapshot); - if (ret) - continue; + struct bpos pagecache_start = bkey_start_pos(&cur->kbuf.k->k); - bch2_btree_iter_set_snapshot(trans, &iter, snapshot); + /* + * Does the pagecache or the btree take precedence? + * + * It _should_ be the pagecache, so that we correctly report delalloc + * extents when dirty in the pagecache (we're COW, after all). + * + * But we'd have to add per-sector writeback tracking to + * bch_folio_state, otherwise we report delalloc extents for clean + * cached data in the pagecache. + * + * We should do this, but even then fiemap won't report stable mappings: + * on bcachefs data moves around in the background (copygc, rebalance) + * and we don't provide a way for userspace to lock that out. + */ + if (k.k && + bkey_le(bpos_max(iter.pos, bkey_start_pos(k.k)), + pagecache_start)) { + bch2_bkey_buf_reassemble(&cur->kbuf, trans->c, k); + bch2_cut_front(iter.pos, cur->kbuf.k); + bch2_cut_back(POS(inode->ei_inum.inum, end), cur->kbuf.k); + cur->flags = 0; + } else if (k.k) { + bch2_cut_back(bkey_start_pos(k.k), cur->kbuf.k); + } - k = bch2_btree_iter_peek_max(trans, &iter, end); - ret = bkey_err(k); + if (cur->kbuf.k->k.type == KEY_TYPE_reflink_p) { + unsigned sectors = cur->kbuf.k->k.size; + s64 offset_into_extent = 0; + enum btree_id data_btree = BTREE_ID_extents; + int ret = bch2_read_indirect_extent(trans, &data_btree, &offset_into_extent, + &cur->kbuf); if (ret) - continue; + goto err; - if (!k.k) - break; + struct bkey_i *k = cur->kbuf.k; + sectors = min_t(unsigned, sectors, k->k.size - offset_into_extent); - if (!bkey_extent_is_data(k.k) && - k.k->type != KEY_TYPE_reservation) { - bch2_btree_iter_advance(trans, &iter); - continue; - } + bch2_cut_front(POS(k->k.p.inode, + bkey_start_offset(&k->k) + offset_into_extent), + k); + bch2_key_resize(&k->k, sectors); + k->k.p = iter.pos; + k->k.p.offset += k->k.size; + } +err: + bch2_trans_iter_exit(trans, &iter); + return ret; +} - s64 offset_into_extent = iter.pos.offset - bkey_start_offset(k.k); - unsigned sectors = k.k->size - offset_into_extent; +static int bch2_fiemap(struct inode *vinode, struct fiemap_extent_info *info, + u64 start, u64 len) +{ + struct bch_fs *c = vinode->i_sb->s_fs_info; + struct bch_inode_info *ei = to_bch_ei(vinode); + struct btree_trans *trans; + struct bch_fiemap_extent cur, prev; + int ret = 0; + + ret = fiemap_prep(&ei->v, info, start, &len, 0); + if (ret) + return ret; + + if (start + len < start) + return -EINVAL; + + start >>= 9; + u64 end = (start + len) >> 9; - bch2_bkey_buf_reassemble(&cur, c, k); + bch2_bkey_buf_init(&cur.kbuf); + bch2_bkey_buf_init(&prev.kbuf); + bkey_init(&prev.kbuf.k->k); - ret = bch2_read_indirect_extent(trans, &data_btree, - &offset_into_extent, &cur); + trans = bch2_trans_get(c); + + while (start < end) { + ret = lockrestart_do(trans, + bch2_next_fiemap_extent(trans, ei, start, end, &cur)); if (ret) - continue; + goto err; - k = bkey_i_to_s_c(cur.k); - bch2_bkey_buf_realloc(&prev, c, k.k->u64s); + BUG_ON(bkey_start_offset(&cur.kbuf.k->k) < start); + BUG_ON(cur.kbuf.k->k.p.offset > end); - sectors = min_t(unsigned, sectors, k.k->size - offset_into_extent); + if (bkey_start_offset(&cur.kbuf.k->k) == end) + break; - bch2_cut_front(POS(k.k->p.inode, - bkey_start_offset(k.k) + - offset_into_extent), - cur.k); - bch2_key_resize(&cur.k->k, sectors); - cur.k->k.p = iter.pos; - cur.k->k.p.offset += cur.k->k.size; + start = cur.kbuf.k->k.p.offset; - if (have_extent) { + if (!bkey_deleted(&prev.kbuf.k->k)) { bch2_trans_unlock(trans); - ret = bch2_fill_extent(c, info, - bkey_i_to_s_c(prev.k), 0); + ret = bch2_fill_extent(c, info, &prev); if (ret) - break; + goto err; } - bkey_copy(prev.k, cur.k); - have_extent = true; - - bch2_btree_iter_set_pos(trans, &iter, - POS(iter.pos.inode, iter.pos.offset + sectors)); + bch2_bkey_buf_copy(&prev.kbuf, c, cur.kbuf.k); + prev.flags = cur.flags; } - bch2_trans_iter_exit(trans, &iter); - if (!ret && have_extent) { + if (!bkey_deleted(&prev.kbuf.k->k)) { bch2_trans_unlock(trans); - ret = bch2_fill_extent(c, info, bkey_i_to_s_c(prev.k), - FIEMAP_EXTENT_LAST); + prev.flags |= FIEMAP_EXTENT_LAST; + ret = bch2_fill_extent(c, info, &prev); } - +err: bch2_trans_put(trans); - bch2_bkey_buf_exit(&cur, c); - bch2_bkey_buf_exit(&prev, c); - return ret < 0 ? ret : 0; + bch2_bkey_buf_exit(&cur.kbuf, c); + bch2_bkey_buf_exit(&prev.kbuf, c); + + return bch2_err_class(ret < 0 ? ret : 0); } static const struct vm_operations_struct bch_vm_ops = { @@ -1449,6 +1599,165 @@ static int bch2_open(struct inode *vinode, struct file *file) return generic_file_open(vinode, file); } +/* bcachefs inode flags -> FS_IOC_GETFLAGS: */ +static const __maybe_unused unsigned bch_flags_to_uflags[] = { + [__BCH_INODE_sync] = FS_SYNC_FL, + [__BCH_INODE_immutable] = FS_IMMUTABLE_FL, + [__BCH_INODE_append] = FS_APPEND_FL, + [__BCH_INODE_nodump] = FS_NODUMP_FL, + [__BCH_INODE_noatime] = FS_NOATIME_FL, +}; + +/* bcachefs inode flags -> FS_IOC_FSGETXATTR: */ +static const __maybe_unused unsigned bch_flags_to_xflags[] = { + [__BCH_INODE_sync] = FS_XFLAG_SYNC, + [__BCH_INODE_immutable] = FS_XFLAG_IMMUTABLE, + [__BCH_INODE_append] = FS_XFLAG_APPEND, + [__BCH_INODE_nodump] = FS_XFLAG_NODUMP, + [__BCH_INODE_noatime] = FS_XFLAG_NOATIME, +}; + +static int bch2_fileattr_get(struct dentry *dentry, + struct fileattr *fa) +{ + struct bch_inode_info *inode = to_bch_ei(d_inode(dentry)); + struct bch_fs *c = inode->v.i_sb->s_fs_info; + + fileattr_fill_xflags(fa, map_flags(bch_flags_to_xflags, inode->ei_inode.bi_flags)); + + if (inode->ei_inode.bi_fields_set & (1 << Inode_opt_project)) + fa->fsx_xflags |= FS_XFLAG_PROJINHERIT; + + if (bch2_inode_casefold(c, &inode->ei_inode)) + fa->flags |= FS_CASEFOLD_FL; + + fa->fsx_projid = inode->ei_qid.q[QTYP_PRJ]; + return 0; +} + +struct flags_set { + unsigned mask; + unsigned flags; + unsigned projid; + bool set_project; + bool set_casefold; + bool casefold; +}; + +static int fssetxattr_inode_update_fn(struct btree_trans *trans, + struct bch_inode_info *inode, + struct bch_inode_unpacked *bi, + void *p) +{ + struct bch_fs *c = trans->c; + struct flags_set *s = p; + + /* + * We're relying on btree locking here for exclusion with other ioctl + * calls - use the flags in the btree (@bi), not inode->i_flags: + */ + if (!S_ISREG(bi->bi_mode) && + !S_ISDIR(bi->bi_mode) && + (s->flags & (BCH_INODE_nodump|BCH_INODE_noatime)) != s->flags) + return -EINVAL; + + if (s->casefold != bch2_inode_casefold(c, bi)) { +#ifdef CONFIG_UNICODE + int ret = 0; + /* Not supported on individual files. */ + if (!S_ISDIR(bi->bi_mode)) + return -EOPNOTSUPP; + + /* + * Make sure the dir is empty, as otherwise we'd need to + * rehash everything and update the dirent keys. + */ + ret = bch2_empty_dir_trans(trans, inode_inum(inode)); + if (ret < 0) + return ret; + + ret = bch2_request_incompat_feature(c, bcachefs_metadata_version_casefolding); + if (ret) + return ret; + + bch2_check_set_feature(c, BCH_FEATURE_casefolding); + + bi->bi_casefold = s->casefold + 1; + bi->bi_fields_set |= BIT(Inode_opt_casefold); + +#else + printk(KERN_ERR "Cannot use casefolding on a kernel without CONFIG_UNICODE\n"); + return -EOPNOTSUPP; +#endif + } + + if (s->set_project) { + bi->bi_project = s->projid; + bi->bi_fields_set |= BIT(Inode_opt_project); + } + + bi->bi_flags &= ~s->mask; + bi->bi_flags |= s->flags; + + bi->bi_ctime = timespec_to_bch2_time(c, current_time(&inode->v)); + return 0; +} + +static int bch2_fileattr_set(struct mnt_idmap *idmap, + struct dentry *dentry, + struct fileattr *fa) +{ + struct bch_inode_info *inode = to_bch_ei(d_inode(dentry)); + struct bch_fs *c = inode->v.i_sb->s_fs_info; + struct flags_set s = {}; + int ret; + + if (fa->fsx_valid) { + fa->fsx_xflags &= ~FS_XFLAG_PROJINHERIT; + + s.mask = map_defined(bch_flags_to_xflags); + s.flags |= map_flags_rev(bch_flags_to_xflags, fa->fsx_xflags); + if (fa->fsx_xflags) + return -EOPNOTSUPP; + + if (fa->fsx_projid >= U32_MAX) + return -EINVAL; + + /* + * inode fields accessible via the xattr interface are stored with a +1 + * bias, so that 0 means unset: + */ + if ((inode->ei_inode.bi_project || + fa->fsx_projid) && + inode->ei_inode.bi_project != fa->fsx_projid + 1) { + s.projid = fa->fsx_projid + 1; + s.set_project = true; + } + } + + if (fa->flags_valid) { + s.mask = map_defined(bch_flags_to_uflags); + + s.set_casefold = true; + s.casefold = (fa->flags & FS_CASEFOLD_FL) != 0; + fa->flags &= ~FS_CASEFOLD_FL; + + s.flags |= map_flags_rev(bch_flags_to_uflags, fa->flags); + if (fa->flags) + return -EOPNOTSUPP; + } + + mutex_lock(&inode->ei_update_lock); + ret = bch2_subvol_is_ro(c, inode->ei_inum.subvol) ?: + (s.set_project + ? bch2_set_projid(c, inode, fa->fsx_projid) + : 0) ?: + bch2_write_inode(c, inode, fssetxattr_inode_update_fn, &s, + ATTR_CTIME); + mutex_unlock(&inode->ei_update_lock); + return ret; +} + static const struct file_operations bch_file_operations = { .open = bch2_open, .llseek = bch2_llseek, @@ -1476,6 +1785,8 @@ static const struct inode_operations bch_file_inode_operations = { .get_inode_acl = bch2_get_acl, .set_acl = bch2_set_acl, #endif + .fileattr_get = bch2_fileattr_get, + .fileattr_set = bch2_fileattr_set, }; static const struct inode_operations bch_dir_inode_operations = { @@ -1496,6 +1807,8 @@ static const struct inode_operations bch_dir_inode_operations = { .get_inode_acl = bch2_get_acl, .set_acl = bch2_set_acl, #endif + .fileattr_get = bch2_fileattr_get, + .fileattr_set = bch2_fileattr_set, }; static const struct file_operations bch_dir_file_operations = { @@ -1518,6 +1831,8 @@ static const struct inode_operations bch_symlink_inode_operations = { .get_inode_acl = bch2_get_acl, .set_acl = bch2_set_acl, #endif + .fileattr_get = bch2_fileattr_get, + .fileattr_set = bch2_fileattr_set, }; static const struct inode_operations bch_special_inode_operations = { @@ -1528,6 +1843,8 @@ static const struct inode_operations bch_special_inode_operations = { .get_inode_acl = bch2_get_acl, .set_acl = bch2_set_acl, #endif + .fileattr_get = bch2_fileattr_get, + .fileattr_set = bch2_fileattr_set, }; static const struct address_space_operations bch_address_space_operations = { diff --git a/fs/bcachefs/inode.h b/fs/bcachefs/inode.h index f82cfbf460d0..c74af15b14f2 100644 --- a/fs/bcachefs/inode.h +++ b/fs/bcachefs/inode.h @@ -243,6 +243,14 @@ static inline unsigned bkey_inode_mode(struct bkey_s_c k) } } +static inline bool bch2_inode_casefold(struct bch_fs *c, const struct bch_inode_unpacked *bi) +{ + /* inode apts are stored with a +1 bias: 0 means "unset, use fs opt" */ + return bi->bi_casefold + ? bi->bi_casefold - 1 + : c->opts.casefold; +} + /* i_nlink: */ static inline unsigned nlink_bias(umode_t mode) diff --git a/fs/bcachefs/inode_format.h b/fs/bcachefs/inode_format.h index 117110af1e3f..87e193e8ed25 100644 --- a/fs/bcachefs/inode_format.h +++ b/fs/bcachefs/inode_format.h @@ -103,7 +103,8 @@ struct bch_inode_generation { x(bi_parent_subvol, 32) \ x(bi_nocow, 8) \ x(bi_depth, 32) \ - x(bi_inodes_32bit, 8) + x(bi_inodes_32bit, 8) \ + x(bi_casefold, 8) /* subset of BCH_INODE_FIELDS */ #define BCH_INODE_OPTS() \ @@ -117,7 +118,8 @@ struct bch_inode_generation { x(background_target, 16) \ x(erasure_code, 16) \ x(nocow, 8) \ - x(inodes_32bit, 8) + x(inodes_32bit, 8) \ + x(casefold, 8) enum inode_opt_id { #define x(name, ...) \ @@ -137,8 +139,7 @@ enum inode_opt_id { x(i_sectors_dirty, 6) \ x(unlinked, 7) \ x(backptr_untrusted, 8) \ - x(has_child_snapshot, 9) \ - x(casefolded, 10) + x(has_child_snapshot, 9) /* bits 20+ reserved for packed fields below: */ diff --git a/fs/bcachefs/journal.c b/fs/bcachefs/journal.c index d8f74b6d0a75..bb45d3634194 100644 --- a/fs/bcachefs/journal.c +++ b/fs/bcachefs/journal.c @@ -281,7 +281,24 @@ static void __journal_entry_close(struct journal *j, unsigned closed_val, bool t sectors = vstruct_blocks_plus(buf->data, c->block_bits, buf->u64s_reserved) << c->block_bits; - BUG_ON(sectors > buf->sectors); + if (unlikely(sectors > buf->sectors)) { + struct printbuf err = PRINTBUF; + err.atomic++; + + prt_printf(&err, "journal entry overran reserved space: %u > %u\n", + sectors, buf->sectors); + prt_printf(&err, "buf u64s %u u64s reserved %u cur_entry_u64s %u block_bits %u\n", + le32_to_cpu(buf->data->u64s), buf->u64s_reserved, + j->cur_entry_u64s, + c->block_bits); + prt_printf(&err, "fatal error - emergency read only"); + bch2_journal_halt_locked(j); + + bch_err(c, "%s", err.buf); + printbuf_exit(&err); + return; + } + buf->sectors = sectors; /* @@ -1462,8 +1479,6 @@ int bch2_fs_journal_start(struct journal *j, u64 cur_seq) j->last_empty_seq = cur_seq - 1; /* to match j->seq */ spin_lock(&j->lock); - - set_bit(JOURNAL_running, &j->flags); j->last_flush_write = jiffies; j->reservations.idx = journal_cur_seq(j); @@ -1474,6 +1489,21 @@ int bch2_fs_journal_start(struct journal *j, u64 cur_seq) return 0; } +void bch2_journal_set_replay_done(struct journal *j) +{ + /* + * journal_space_available must happen before setting JOURNAL_running + * JOURNAL_running must happen before JOURNAL_replay_done + */ + spin_lock(&j->lock); + bch2_journal_space_available(j); + + set_bit(JOURNAL_need_flush_write, &j->flags); + set_bit(JOURNAL_running, &j->flags); + set_bit(JOURNAL_replay_done, &j->flags); + spin_unlock(&j->lock); +} + /* init/exit: */ void bch2_dev_journal_exit(struct bch_dev *ca) diff --git a/fs/bcachefs/journal.h b/fs/bcachefs/journal.h index 47828771f9c2..641e20c05a14 100644 --- a/fs/bcachefs/journal.h +++ b/fs/bcachefs/journal.h @@ -437,12 +437,6 @@ static inline int bch2_journal_error(struct journal *j) struct bch_dev; -static inline void bch2_journal_set_replay_done(struct journal *j) -{ - BUG_ON(!test_bit(JOURNAL_running, &j->flags)); - set_bit(JOURNAL_replay_done, &j->flags); -} - void bch2_journal_unblock(struct journal *); void bch2_journal_block(struct journal *); struct journal_buf *bch2_next_write_buffer_flush_journal_buf(struct journal *, u64, bool *); @@ -459,6 +453,7 @@ void bch2_dev_journal_stop(struct journal *, struct bch_dev *); void bch2_fs_journal_stop(struct journal *); int bch2_fs_journal_start(struct journal *, u64); +void bch2_journal_set_replay_done(struct journal *); void bch2_dev_journal_exit(struct bch_dev *); int bch2_dev_journal_init(struct bch_dev *, struct bch_sb *); diff --git a/fs/bcachefs/journal_reclaim.c b/fs/bcachefs/journal_reclaim.c index 5d1547aa118a..ea670c3c43d8 100644 --- a/fs/bcachefs/journal_reclaim.c +++ b/fs/bcachefs/journal_reclaim.c @@ -252,7 +252,10 @@ void bch2_journal_space_available(struct journal *j) bch2_journal_set_watermark(j); out: - j->cur_entry_sectors = !ret ? j->space[journal_space_discarded].next_entry : 0; + j->cur_entry_sectors = !ret + ? round_down(j->space[journal_space_discarded].next_entry, + block_sectors(c)) + : 0; j->cur_entry_error = ret; if (!ret) diff --git a/fs/bcachefs/movinggc.c b/fs/bcachefs/movinggc.c index 159410c50861..96873372b516 100644 --- a/fs/bcachefs/movinggc.c +++ b/fs/bcachefs/movinggc.c @@ -356,6 +356,13 @@ static int bch2_copygc_thread(void *arg) set_freezable(); + /* + * Data move operations can't run until after check_snapshots has + * completed, and bch2_snapshot_is_ancestor() is available. + */ + kthread_wait_freezable(c->recovery_pass_done > BCH_RECOVERY_PASS_check_snapshots || + kthread_should_stop()); + bch2_move_stats_init(&move_stats, "copygc"); bch2_moving_ctxt_init(&ctxt, c, NULL, &move_stats, writepoint_ptr(&c->copygc_write_point), diff --git a/fs/bcachefs/movinggc.h b/fs/bcachefs/movinggc.h index ea181fef5bc9..d1885cf67a45 100644 --- a/fs/bcachefs/movinggc.h +++ b/fs/bcachefs/movinggc.h @@ -5,6 +5,15 @@ unsigned long bch2_copygc_wait_amount(struct bch_fs *); void bch2_copygc_wait_to_text(struct printbuf *, struct bch_fs *); +static inline void bch2_copygc_wakeup(struct bch_fs *c) +{ + rcu_read_lock(); + struct task_struct *p = rcu_dereference(c->copygc_thread); + if (p) + wake_up_process(p); + rcu_read_unlock(); +} + void bch2_copygc_stop(struct bch_fs *); int bch2_copygc_start(struct bch_fs *); void bch2_fs_copygc_init(struct bch_fs *); diff --git a/fs/bcachefs/namei.c b/fs/bcachefs/namei.c index 0d65ea96f7a2..46f3c8b100a9 100644 --- a/fs/bcachefs/namei.c +++ b/fs/bcachefs/namei.c @@ -47,10 +47,6 @@ int bch2_create_trans(struct btree_trans *trans, if (ret) goto err; - /* Inherit casefold state from parent. */ - if (S_ISDIR(mode)) - new_inode->bi_flags |= dir_u->bi_flags & BCH_INODE_casefolded; - if (!(flags & BCH_CREATE_SNAPSHOT)) { /* Normal create path - allocate a new inode: */ bch2_inode_init_late(new_inode, now, uid, gid, mode, rdev, dir_u); diff --git a/fs/bcachefs/opts.h b/fs/bcachefs/opts.h index 4d06313076ff..dfb14810124c 100644 --- a/fs/bcachefs/opts.h +++ b/fs/bcachefs/opts.h @@ -228,6 +228,11 @@ enum fsck_err_opts { OPT_BOOL(), \ BCH_SB_ERASURE_CODE, false, \ NULL, "Enable erasure coding (DO NOT USE YET)") \ + x(casefold, u8, \ + OPT_FS|OPT_INODE|OPT_FORMAT, \ + OPT_BOOL(), \ + BCH_SB_CASEFOLD, false, \ + NULL, "Dirent lookups are casefolded") \ x(inodes_32bit, u8, \ OPT_FS|OPT_INODE|OPT_FORMAT|OPT_MOUNT|OPT_RUNTIME, \ OPT_BOOL(), \ diff --git a/fs/bcachefs/rebalance.c b/fs/bcachefs/rebalance.c index c63fa53f30d2..4ccdfc1f34aa 100644 --- a/fs/bcachefs/rebalance.c +++ b/fs/bcachefs/rebalance.c @@ -262,7 +262,7 @@ int bch2_set_rebalance_needs_scan(struct bch_fs *c, u64 inum) int ret = bch2_trans_commit_do(c, NULL, NULL, BCH_TRANS_COMMIT_no_enospc, bch2_set_rebalance_needs_scan_trans(trans, inum)); - rebalance_wakeup(c); + bch2_rebalance_wakeup(c); return ret; } @@ -581,6 +581,13 @@ static int bch2_rebalance_thread(void *arg) set_freezable(); + /* + * Data move operations can't run until after check_snapshots has + * completed, and bch2_snapshot_is_ancestor() is available. + */ + kthread_wait_freezable(c->recovery_pass_done > BCH_RECOVERY_PASS_check_snapshots || + kthread_should_stop()); + bch2_moving_ctxt_init(&ctxt, c, NULL, &r->work_stats, writepoint_ptr(&c->rebalance_write_point), true); @@ -664,7 +671,7 @@ void bch2_rebalance_stop(struct bch_fs *c) c->rebalance.thread = NULL; if (p) { - /* for sychronizing with rebalance_wakeup() */ + /* for sychronizing with bch2_rebalance_wakeup() */ synchronize_rcu(); kthread_stop(p); diff --git a/fs/bcachefs/rebalance.h b/fs/bcachefs/rebalance.h index 62a3859d3823..e5e8eb4a2dd1 100644 --- a/fs/bcachefs/rebalance.h +++ b/fs/bcachefs/rebalance.h @@ -37,7 +37,7 @@ int bch2_set_rebalance_needs_scan_trans(struct btree_trans *, u64); int bch2_set_rebalance_needs_scan(struct bch_fs *, u64 inum); int bch2_set_fs_needs_rebalance(struct bch_fs *); -static inline void rebalance_wakeup(struct bch_fs *c) +static inline void bch2_rebalance_wakeup(struct bch_fs *c) { struct task_struct *p; diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c index 606d684e6f23..d6c4ef819d40 100644 --- a/fs/bcachefs/recovery.c +++ b/fs/bcachefs/recovery.c @@ -18,6 +18,7 @@ #include "journal_seq_blacklist.h" #include "logged_ops.h" #include "move.h" +#include "movinggc.h" #include "namei.h" #include "quota.h" #include "rebalance.h" @@ -1129,13 +1130,13 @@ int bch2_fs_initialize(struct bch_fs *c) if (ret) goto err; - set_bit(BCH_FS_accounting_replay_done, &c->flags); - bch2_journal_set_replay_done(&c->journal); - ret = bch2_fs_read_write_early(c); if (ret) goto err; + set_bit(BCH_FS_accounting_replay_done, &c->flags); + bch2_journal_set_replay_done(&c->journal); + for_each_member_device(c, ca) { ret = bch2_dev_usage_init(ca, false); if (ret) { @@ -1194,6 +1195,9 @@ int bch2_fs_initialize(struct bch_fs *c) c->recovery_pass_done = BCH_RECOVERY_PASS_NR - 1; + bch2_copygc_wakeup(c); + bch2_rebalance_wakeup(c); + if (enabled_qtypes(c)) { ret = bch2_fs_quota_read(c); if (ret) diff --git a/fs/bcachefs/recovery_passes.c b/fs/bcachefs/recovery_passes.c index 593ff142530d..22f72bb5b853 100644 --- a/fs/bcachefs/recovery_passes.c +++ b/fs/bcachefs/recovery_passes.c @@ -12,6 +12,7 @@ #include "journal.h" #include "lru.h" #include "logged_ops.h" +#include "movinggc.h" #include "rebalance.h" #include "recovery.h" #include "recovery_passes.h" @@ -262,49 +263,52 @@ int bch2_run_recovery_passes(struct bch_fs *c) */ c->opts.recovery_passes_exclude &= ~BCH_RECOVERY_PASS_set_may_go_rw; - while (c->curr_recovery_pass < ARRAY_SIZE(recovery_pass_fns) && !ret) { - c->next_recovery_pass = c->curr_recovery_pass + 1; + spin_lock_irq(&c->recovery_pass_lock); - spin_lock_irq(&c->recovery_pass_lock); + while (c->curr_recovery_pass < ARRAY_SIZE(recovery_pass_fns) && !ret) { + unsigned prev_done = c->recovery_pass_done; unsigned pass = c->curr_recovery_pass; + c->next_recovery_pass = pass + 1; + if (c->opts.recovery_pass_last && - c->curr_recovery_pass > c->opts.recovery_pass_last) { - spin_unlock_irq(&c->recovery_pass_lock); + c->curr_recovery_pass > c->opts.recovery_pass_last) break; - } - if (!should_run_recovery_pass(c, pass)) { - c->curr_recovery_pass++; - c->recovery_pass_done = max(c->recovery_pass_done, pass); + if (should_run_recovery_pass(c, pass)) { spin_unlock_irq(&c->recovery_pass_lock); - continue; - } - spin_unlock_irq(&c->recovery_pass_lock); - - ret = bch2_run_recovery_pass(c, pass) ?: - bch2_journal_flush(&c->journal); - - if (!ret && !test_bit(BCH_FS_error, &c->flags)) - bch2_clear_recovery_pass_required(c, pass); - - spin_lock_irq(&c->recovery_pass_lock); - if (c->next_recovery_pass < c->curr_recovery_pass) { - /* - * bch2_run_explicit_recovery_pass() was called: we - * can't always catch -BCH_ERR_restart_recovery because - * it may have been called from another thread (btree - * node read completion) - */ - ret = 0; - c->recovery_passes_complete &= ~(~0ULL << c->curr_recovery_pass); - } else { - c->recovery_passes_complete |= BIT_ULL(pass); - c->recovery_pass_done = max(c->recovery_pass_done, pass); + ret = bch2_run_recovery_pass(c, pass) ?: + bch2_journal_flush(&c->journal); + + if (!ret && !test_bit(BCH_FS_error, &c->flags)) + bch2_clear_recovery_pass_required(c, pass); + spin_lock_irq(&c->recovery_pass_lock); + + if (c->next_recovery_pass < c->curr_recovery_pass) { + /* + * bch2_run_explicit_recovery_pass() was called: we + * can't always catch -BCH_ERR_restart_recovery because + * it may have been called from another thread (btree + * node read completion) + */ + ret = 0; + c->recovery_passes_complete &= ~(~0ULL << c->curr_recovery_pass); + } else { + c->recovery_passes_complete |= BIT_ULL(pass); + c->recovery_pass_done = max(c->recovery_pass_done, pass); + } } + c->curr_recovery_pass = c->next_recovery_pass; - spin_unlock_irq(&c->recovery_pass_lock); + + if (prev_done <= BCH_RECOVERY_PASS_check_snapshots && + c->recovery_pass_done > BCH_RECOVERY_PASS_check_snapshots) { + bch2_copygc_wakeup(c); + bch2_rebalance_wakeup(c); + } } + spin_unlock_irq(&c->recovery_pass_lock); + return ret; } diff --git a/fs/bcachefs/snapshot.c b/fs/bcachefs/snapshot.c index b7de29aed839..fec569c7deb1 100644 --- a/fs/bcachefs/snapshot.c +++ b/fs/bcachefs/snapshot.c @@ -396,7 +396,7 @@ u32 bch2_snapshot_tree_oldest_subvol(struct bch_fs *c, u32 snapshot_root) u32 subvol = 0, s; rcu_read_lock(); - while (id) { + while (id && bch2_snapshot_exists(c, id)) { s = snapshot_t(c, id)->subvol; if (s && (!subvol || s < subvol)) diff --git a/fs/bcachefs/str_hash.h b/fs/bcachefs/str_hash.h index 09a354a26c3b..0c1a00539bd1 100644 --- a/fs/bcachefs/str_hash.h +++ b/fs/bcachefs/str_hash.h @@ -33,7 +33,7 @@ bch2_str_hash_opt_to_type(struct bch_fs *c, enum bch_str_hash_opts opt) struct bch_hash_info { u8 type; - struct unicode_map *cf_encoding; + struct unicode_map *cf_encoding; /* * For crc32 or crc64 string hashes the first key value of * the siphash_key (k0) is used as the key. @@ -44,11 +44,10 @@ struct bch_hash_info { static inline struct bch_hash_info bch2_hash_info_init(struct bch_fs *c, const struct bch_inode_unpacked *bi) { - /* XXX ick */ struct bch_hash_info info = { .type = INODE_STR_HASH(bi), #ifdef CONFIG_UNICODE - .cf_encoding = !!(bi->bi_flags & BCH_INODE_casefolded) ? c->cf_encoding : NULL, + .cf_encoding = bch2_inode_casefold(c, bi) ? c->cf_encoding : NULL, #endif .siphash_key = { .k0 = bi->bi_hash_seed } }; diff --git a/fs/bcachefs/super-io.c b/fs/bcachefs/super-io.c index 25b6bce05c3c..cb5d960aed92 100644 --- a/fs/bcachefs/super-io.c +++ b/fs/bcachefs/super-io.c @@ -1102,7 +1102,8 @@ int bch2_write_super(struct bch_fs *c) prt_str(&buf, ")"); bch2_fs_fatal_error(c, ": %s", buf.buf); printbuf_exit(&buf); - return -BCH_ERR_sb_not_downgraded; + ret = -BCH_ERR_sb_not_downgraded; + goto out; } darray_for_each(online_devices, ca) { diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c index e8a17ed1615d..e4ab0595c0ae 100644 --- a/fs/bcachefs/super.c +++ b/fs/bcachefs/super.c @@ -418,32 +418,6 @@ bool bch2_fs_emergency_read_only_locked(struct bch_fs *c) return ret; } -static int bch2_fs_read_write_late(struct bch_fs *c) -{ - int ret; - - /* - * Data move operations can't run until after check_snapshots has - * completed, and bch2_snapshot_is_ancestor() is available. - * - * Ideally we'd start copygc/rebalance earlier instead of waiting for - * all of recovery/fsck to complete: - */ - ret = bch2_copygc_start(c); - if (ret) { - bch_err(c, "error starting copygc thread"); - return ret; - } - - ret = bch2_rebalance_start(c); - if (ret) { - bch_err(c, "error starting rebalance thread"); - return ret; - } - - return 0; -} - static int __bch2_fs_read_write(struct bch_fs *c, bool early) { int ret; @@ -466,29 +440,28 @@ static int __bch2_fs_read_write(struct bch_fs *c, bool early) clear_bit(BCH_FS_clean_shutdown, &c->flags); + __for_each_online_member(c, ca, BIT(BCH_MEMBER_STATE_rw), READ) { + bch2_dev_allocator_add(c, ca); + percpu_ref_reinit(&ca->io_ref[WRITE]); + } + bch2_recalc_capacity(c); + /* * First journal write must be a flush write: after a clean shutdown we * don't read the journal, so the first journal write may end up * overwriting whatever was there previously, and there must always be * at least one non-flush write in the journal or recovery will fail: */ + spin_lock(&c->journal.lock); set_bit(JOURNAL_need_flush_write, &c->journal.flags); set_bit(JOURNAL_running, &c->journal.flags); - - __for_each_online_member(c, ca, BIT(BCH_MEMBER_STATE_rw), READ) { - bch2_dev_allocator_add(c, ca); - percpu_ref_reinit(&ca->io_ref[WRITE]); - } - bch2_recalc_capacity(c); + bch2_journal_space_available(&c->journal); + spin_unlock(&c->journal.lock); ret = bch2_fs_mark_dirty(c); if (ret) goto err; - spin_lock(&c->journal.lock); - bch2_journal_space_available(&c->journal); - spin_unlock(&c->journal.lock); - ret = bch2_journal_reclaim_start(&c->journal); if (ret) goto err; @@ -504,10 +477,17 @@ static int __bch2_fs_read_write(struct bch_fs *c, bool early) atomic_long_inc(&c->writes[i]); } #endif - if (!early) { - ret = bch2_fs_read_write_late(c); - if (ret) - goto err; + + ret = bch2_copygc_start(c); + if (ret) { + bch_err_msg(c, ret, "error starting copygc thread"); + goto err; + } + + ret = bch2_rebalance_start(c); + if (ret) { + bch_err_msg(c, ret, "error starting rebalance thread"); + goto err; } bch2_do_discards(c); @@ -553,6 +533,7 @@ static void __bch2_fs_free(struct bch_fs *c) bch2_find_btree_nodes_exit(&c->found_btree_nodes); bch2_free_pending_node_rewrites(c); + bch2_free_fsck_errs(c); bch2_fs_accounting_exit(c); bch2_fs_sb_errors_exit(c); bch2_fs_counters_exit(c); @@ -1023,6 +1004,40 @@ static void print_mount_opts(struct bch_fs *c) printbuf_exit(&p); } +static bool bch2_fs_may_start(struct bch_fs *c) +{ + struct bch_dev *ca; + unsigned i, flags = 0; + + if (c->opts.very_degraded) + flags |= BCH_FORCE_IF_DEGRADED|BCH_FORCE_IF_LOST; + + if (c->opts.degraded) + flags |= BCH_FORCE_IF_DEGRADED; + + if (!c->opts.degraded && + !c->opts.very_degraded) { + mutex_lock(&c->sb_lock); + + for (i = 0; i < c->disk_sb.sb->nr_devices; i++) { + if (!bch2_member_exists(c->disk_sb.sb, i)) + continue; + + ca = bch2_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); + } + + return bch2_have_enough_devs(c, bch2_online_devs(c), flags, true); +} + int bch2_fs_start(struct bch_fs *c) { time64_t now = ktime_get_real_seconds(); @@ -1030,6 +1045,9 @@ int bch2_fs_start(struct bch_fs *c) print_mount_opts(c); + if (!bch2_fs_may_start(c)) + return -BCH_ERR_insufficient_devices_to_start; + down_write(&c->state_lock); mutex_lock(&c->sb_lock); @@ -1082,13 +1100,10 @@ int bch2_fs_start(struct bch_fs *c) wake_up(&c->ro_ref_wait); down_write(&c->state_lock); - if (c->opts.read_only) { + if (c->opts.read_only) bch2_fs_read_only(c); - } else { - ret = !test_bit(BCH_FS_rw, &c->flags) - ? bch2_fs_read_write(c) - : bch2_fs_read_write_late(c); - } + else if (!test_bit(BCH_FS_rw, &c->flags)) + ret = bch2_fs_read_write(c); up_write(&c->state_lock); err: @@ -1500,7 +1515,7 @@ static int bch2_dev_attach_bdev(struct bch_fs *c, struct bch_sb_handle *sb) printbuf_exit(&name); - rebalance_wakeup(c); + bch2_rebalance_wakeup(c); return 0; } @@ -1559,40 +1574,6 @@ bool bch2_dev_state_allowed(struct bch_fs *c, struct bch_dev *ca, } } -static bool bch2_fs_may_start(struct bch_fs *c) -{ - struct bch_dev *ca; - unsigned i, flags = 0; - - if (c->opts.very_degraded) - flags |= BCH_FORCE_IF_DEGRADED|BCH_FORCE_IF_LOST; - - if (c->opts.degraded) - flags |= BCH_FORCE_IF_DEGRADED; - - if (!c->opts.degraded && - !c->opts.very_degraded) { - mutex_lock(&c->sb_lock); - - for (i = 0; i < c->disk_sb.sb->nr_devices; i++) { - if (!bch2_member_exists(c->disk_sb.sb, i)) - continue; - - ca = bch2_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); - } - - return bch2_have_enough_devs(c, bch2_online_devs(c), flags, true); -} - static void __bch2_dev_read_only(struct bch_fs *c, struct bch_dev *ca) { bch2_dev_io_ref_stop(ca, WRITE); @@ -1646,7 +1627,7 @@ int __bch2_dev_set_state(struct bch_fs *c, struct bch_dev *ca, if (new_state == BCH_MEMBER_STATE_rw) __bch2_dev_read_write(c, ca); - rebalance_wakeup(c); + bch2_rebalance_wakeup(c); return ret; } @@ -2228,11 +2209,6 @@ struct bch_fs *bch2_fs_open(char * const *devices, unsigned nr_devices, } up_write(&c->state_lock); - if (!bch2_fs_may_start(c)) { - ret = -BCH_ERR_insufficient_devices_to_start; - goto err_print; - } - if (!c->opts.nostart) { ret = bch2_fs_start(c); if (ret) diff --git a/fs/bcachefs/sysfs.c b/fs/bcachefs/sysfs.c index e5f003c29369..82ee333ddd21 100644 --- a/fs/bcachefs/sysfs.c +++ b/fs/bcachefs/sysfs.c @@ -654,11 +654,10 @@ static ssize_t sysfs_opt_store(struct bch_fs *c, bch2_set_rebalance_needs_scan(c, 0); if (v && id == Opt_rebalance_enabled) - rebalance_wakeup(c); + bch2_rebalance_wakeup(c); - if (v && id == Opt_copygc_enabled && - c->copygc_thread) - wake_up_process(c->copygc_thread); + if (v && id == Opt_copygc_enabled) + bch2_copygc_wakeup(c); if (id == Opt_discard && !ca) { mutex_lock(&c->sb_lock); diff --git a/fs/bcachefs/tests.c b/fs/bcachefs/tests.c index c265b102267a..782a05fe7656 100644 --- a/fs/bcachefs/tests.c +++ b/fs/bcachefs/tests.c @@ -342,6 +342,8 @@ static int test_iterate_slots_extents(struct bch_fs *c, u64 nr) */ static int test_peek_end(struct bch_fs *c, u64 nr) { + delete_test_keys(c); + struct btree_trans *trans = bch2_trans_get(c); struct btree_iter iter; struct bkey_s_c k; @@ -362,6 +364,8 @@ static int test_peek_end(struct bch_fs *c, u64 nr) static int test_peek_end_extents(struct bch_fs *c, u64 nr) { + delete_test_keys(c); + struct btree_trans *trans = bch2_trans_get(c); struct btree_iter iter; struct bkey_s_c k; diff --git a/fs/bcachefs/util.h b/fs/bcachefs/util.h index 6ba5071ab6dd..3e52c7f8ddd2 100644 --- a/fs/bcachefs/util.h +++ b/fs/bcachefs/util.h @@ -739,4 +739,42 @@ static inline void memcpy_swab(void *_dst, void *_src, size_t len) *--dst = *src++; } +#define set_flags(_map, _in, _out) \ +do { \ + unsigned _i; \ + \ + for (_i = 0; _i < ARRAY_SIZE(_map); _i++) \ + if ((_in) & (1 << _i)) \ + (_out) |= _map[_i]; \ + else \ + (_out) &= ~_map[_i]; \ +} while (0) + +#define map_flags(_map, _in) \ +({ \ + unsigned _out = 0; \ + \ + set_flags(_map, _in, _out); \ + _out; \ +}) + +#define map_flags_rev(_map, _in) \ +({ \ + unsigned _i, _out = 0; \ + \ + for (_i = 0; _i < ARRAY_SIZE(_map); _i++) \ + if ((_in) & _map[_i]) { \ + (_out) |= 1 << _i; \ + (_in) &= ~_map[_i]; \ + } \ + (_out); \ +}) + +#define map_defined(_map) \ +({ \ + unsigned _in = ~0; \ + \ + map_flags_rev(_map, _in); \ +}) + #endif /* _BCACHEFS_UTIL_H */ diff --git a/fs/buffer.c b/fs/buffer.c index c7abb4a029dc..7be23ff20b27 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -176,18 +176,8 @@ void end_buffer_write_sync(struct buffer_head *bh, int uptodate) } EXPORT_SYMBOL(end_buffer_write_sync); -/* - * Various filesystems appear to want __find_get_block to be non-blocking. - * But it's the page lock which protects the buffers. To get around this, - * we get exclusion from try_to_free_buffers with the blockdev mapping's - * i_private_lock. - * - * Hack idea: for the blockdev mapping, i_private_lock contention - * may be quite high. This code could TryLock the page, and if that - * succeeds, there is no need to take i_private_lock. - */ static struct buffer_head * -__find_get_block_slow(struct block_device *bdev, sector_t block) +__find_get_block_slow(struct block_device *bdev, sector_t block, bool atomic) { struct address_space *bd_mapping = bdev->bd_mapping; const int blkbits = bd_mapping->host->i_blkbits; @@ -204,10 +194,28 @@ __find_get_block_slow(struct block_device *bdev, sector_t block) if (IS_ERR(folio)) goto out; - spin_lock(&bd_mapping->i_private_lock); + /* + * Folio lock protects the buffers. Callers that cannot block + * will fallback to serializing vs try_to_free_buffers() via + * the i_private_lock. + */ + if (atomic) + spin_lock(&bd_mapping->i_private_lock); + else + folio_lock(folio); + head = folio_buffers(folio); if (!head) goto out_unlock; + /* + * Upon a noref migration, the folio lock serializes here; + * otherwise bail. + */ + if (test_bit_acquire(BH_Migrate, &head->b_state)) { + WARN_ON(!atomic); + goto out_unlock; + } + bh = head; do { if (!buffer_mapped(bh)) @@ -236,7 +244,10 @@ __find_get_block_slow(struct block_device *bdev, sector_t block) 1 << blkbits); } out_unlock: - spin_unlock(&bd_mapping->i_private_lock); + if (atomic) + spin_unlock(&bd_mapping->i_private_lock); + else + folio_unlock(folio); folio_put(folio); out: return ret; @@ -656,7 +667,9 @@ EXPORT_SYMBOL(generic_buffers_fsync); void write_boundary_block(struct block_device *bdev, sector_t bblock, unsigned blocksize) { - struct buffer_head *bh = __find_get_block(bdev, bblock + 1, blocksize); + struct buffer_head *bh; + + bh = __find_get_block_nonatomic(bdev, bblock + 1, blocksize); if (bh) { if (buffer_dirty(bh)) write_dirty_buffer(bh, 0); @@ -1386,16 +1399,18 @@ lookup_bh_lru(struct block_device *bdev, sector_t block, unsigned size) /* * Perform a pagecache lookup for the matching buffer. If it's there, refresh * it in the LRU and mark it as accessed. If it is not present then return - * NULL + * NULL. Atomic context callers may also return NULL if the buffer is being + * migrated; similarly the page is not marked accessed either. */ -struct buffer_head * -__find_get_block(struct block_device *bdev, sector_t block, unsigned size) +static struct buffer_head * +find_get_block_common(struct block_device *bdev, sector_t block, + unsigned size, bool atomic) { struct buffer_head *bh = lookup_bh_lru(bdev, block, size); if (bh == NULL) { /* __find_get_block_slow will mark the page accessed */ - bh = __find_get_block_slow(bdev, block); + bh = __find_get_block_slow(bdev, block, atomic); if (bh) bh_lru_install(bh); } else @@ -1403,8 +1418,23 @@ __find_get_block(struct block_device *bdev, sector_t block, unsigned size) return bh; } + +struct buffer_head * +__find_get_block(struct block_device *bdev, sector_t block, unsigned size) +{ + return find_get_block_common(bdev, block, size, true); +} EXPORT_SYMBOL(__find_get_block); +/* same as __find_get_block() but allows sleeping contexts */ +struct buffer_head * +__find_get_block_nonatomic(struct block_device *bdev, sector_t block, + unsigned size) +{ + return find_get_block_common(bdev, block, size, false); +} +EXPORT_SYMBOL(__find_get_block_nonatomic); + /** * bdev_getblk - Get a buffer_head in a block device's buffer cache. * @bdev: The block device. @@ -1422,7 +1452,12 @@ EXPORT_SYMBOL(__find_get_block); struct buffer_head *bdev_getblk(struct block_device *bdev, sector_t block, unsigned size, gfp_t gfp) { - struct buffer_head *bh = __find_get_block(bdev, block, size); + struct buffer_head *bh; + + if (gfpflags_allow_blocking(gfp)) + bh = __find_get_block_nonatomic(bdev, block, size); + else + bh = __find_get_block(bdev, block, size); might_alloc(gfp); if (bh) diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index 6ac2bd555e86..06cd2963e41e 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -2367,7 +2367,7 @@ static int fill_fscrypt_truncate(struct inode *inode, /* Try to writeback the dirty pagecaches */ if (issued & (CEPH_CAP_FILE_BUFFER)) { - loff_t lend = orig_pos + CEPH_FSCRYPT_BLOCK_SHIFT - 1; + loff_t lend = orig_pos + CEPH_FSCRYPT_BLOCK_SIZE - 1; ret = filemap_write_and_wait_range(inode->i_mapping, orig_pos, lend); diff --git a/fs/ext4/ialloc.c b/fs/ext4/ialloc.c index 38bc8d74f4cc..e7ecc7c8a729 100644 --- a/fs/ext4/ialloc.c +++ b/fs/ext4/ialloc.c @@ -691,7 +691,8 @@ static int recently_deleted(struct super_block *sb, ext4_group_t group, int ino) if (!bh || !buffer_uptodate(bh)) /* * If the block is not in the buffer cache, then it - * must have been written out. + * must have been written out, or, most unlikely, is + * being migrated - false failure should be OK here. */ goto out; diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index f88424c28194..1e98c5be4e0a 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -6642,7 +6642,8 @@ void ext4_free_blocks(handle_t *handle, struct inode *inode, for (i = 0; i < count; i++) { cond_resched(); if (is_metadata) - bh = sb_find_get_block(inode->i_sb, block + i); + bh = sb_find_get_block_nonatomic(inode->i_sb, + block + i); ext4_forget(handle, is_metadata, inode, bh, block + i); } } diff --git a/fs/file.c b/fs/file.c index dc3f7e120e3e..3a3146664cf3 100644 --- a/fs/file.c +++ b/fs/file.c @@ -26,7 +26,7 @@ #include "internal.h" -bool __file_ref_put_badval(file_ref_t *ref, unsigned long cnt) +static noinline bool __file_ref_put_badval(file_ref_t *ref, unsigned long cnt) { /* * If the reference count was already in the dead zone, then this diff --git a/fs/jbd2/revoke.c b/fs/jbd2/revoke.c index 0cf0fddbee81..1467f6790747 100644 --- a/fs/jbd2/revoke.c +++ b/fs/jbd2/revoke.c @@ -345,7 +345,8 @@ int jbd2_journal_revoke(handle_t *handle, unsigned long long blocknr, bh = bh_in; if (!bh) { - bh = __find_get_block(bdev, blocknr, journal->j_blocksize); + bh = __find_get_block_nonatomic(bdev, blocknr, + journal->j_blocksize); if (bh) BUFFER_TRACE(bh, "found on hash"); } @@ -355,7 +356,8 @@ int jbd2_journal_revoke(handle_t *handle, unsigned long long blocknr, /* If there is a different buffer_head lying around in * memory anywhere... */ - bh2 = __find_get_block(bdev, blocknr, journal->j_blocksize); + bh2 = __find_get_block_nonatomic(bdev, blocknr, + journal->j_blocksize); if (bh2) { /* ... and it has RevokeValid status... */ if (bh2 != bh && buffer_revokevalid(bh2)) @@ -464,7 +466,8 @@ void jbd2_journal_cancel_revoke(handle_t *handle, struct journal_head *jh) * state machine will get very upset later on. */ if (need_cancel) { struct buffer_head *bh2; - bh2 = __find_get_block(bh->b_bdev, bh->b_blocknr, bh->b_size); + bh2 = __find_get_block_nonatomic(bh->b_bdev, bh->b_blocknr, + bh->b_size); if (bh2) { if (bh2 != bh) clear_buffer_revoked(bh2); @@ -492,9 +495,9 @@ void jbd2_clear_buffer_revoked_flags(journal_t *journal) struct jbd2_revoke_record_s *record; struct buffer_head *bh; record = (struct jbd2_revoke_record_s *)list_entry; - bh = __find_get_block(journal->j_fs_dev, - record->blocknr, - journal->j_blocksize); + bh = __find_get_block_nonatomic(journal->j_fs_dev, + record->blocknr, + journal->j_blocksize); if (bh) { clear_buffer_revoked(bh); __brelse(bh); diff --git a/fs/namespace.c b/fs/namespace.c index d9ca80dcc544..98a5cd756e9a 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -2826,56 +2826,62 @@ static struct mountpoint *do_lock_mount(struct path *path, bool beneath) struct vfsmount *mnt = path->mnt; struct dentry *dentry; struct mountpoint *mp = ERR_PTR(-ENOENT); + struct path under = {}; for (;;) { - struct mount *m; + struct mount *m = real_mount(mnt); if (beneath) { - m = real_mount(mnt); + path_put(&under); read_seqlock_excl(&mount_lock); - dentry = dget(m->mnt_mountpoint); + under.mnt = mntget(&m->mnt_parent->mnt); + under.dentry = dget(m->mnt_mountpoint); read_sequnlock_excl(&mount_lock); + dentry = under.dentry; } else { dentry = path->dentry; } inode_lock(dentry->d_inode); - if (unlikely(cant_mount(dentry))) { - inode_unlock(dentry->d_inode); - goto out; - } - namespace_lock(); - if (beneath && (!is_mounted(mnt) || m->mnt_mountpoint != dentry)) { + if (unlikely(cant_mount(dentry) || !is_mounted(mnt))) + break; // not to be mounted on + + if (beneath && unlikely(m->mnt_mountpoint != dentry || + &m->mnt_parent->mnt != under.mnt)) { namespace_unlock(); inode_unlock(dentry->d_inode); - goto out; + continue; // got moved } mnt = lookup_mnt(path); - if (likely(!mnt)) + if (unlikely(mnt)) { + namespace_unlock(); + inode_unlock(dentry->d_inode); + path_put(path); + path->mnt = mnt; + path->dentry = dget(mnt->mnt_root); + continue; // got overmounted + } + mp = get_mountpoint(dentry); + if (IS_ERR(mp)) break; - - namespace_unlock(); - inode_unlock(dentry->d_inode); - if (beneath) - dput(dentry); - path_put(path); - path->mnt = mnt; - path->dentry = dget(mnt->mnt_root); - } - - mp = get_mountpoint(dentry); - if (IS_ERR(mp)) { - namespace_unlock(); - inode_unlock(dentry->d_inode); + if (beneath) { + /* + * @under duplicates the references that will stay + * at least until namespace_unlock(), so the path_put() + * below is safe (and OK to do under namespace_lock - + * we are not dropping the final references here). + */ + path_put(&under); + } + return mp; } - -out: + namespace_unlock(); + inode_unlock(dentry->d_inode); if (beneath) - dput(dentry); - + path_put(&under); return mp; } @@ -2886,14 +2892,11 @@ static inline struct mountpoint *lock_mount(struct path *path) static void unlock_mount(struct mountpoint *where) { - struct dentry *dentry = where->m_dentry; - + inode_unlock(where->m_dentry->d_inode); read_seqlock_excl(&mount_lock); put_mountpoint(where); read_sequnlock_excl(&mount_lock); - namespace_unlock(); - inode_unlock(dentry->d_inode); } static int graft_tree(struct mount *mnt, struct mount *p, struct mountpoint *mp) diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c index f1b4b3e611cb..c7a9729dc9d0 100644 --- a/fs/ocfs2/journal.c +++ b/fs/ocfs2/journal.c @@ -1249,7 +1249,7 @@ static int ocfs2_force_read_journal(struct inode *inode) } for (i = 0; i < p_blocks; i++, p_blkno++) { - bh = __find_get_block(osb->sb->s_bdev, p_blkno, + bh = __find_get_block_nonatomic(osb->sb->s_bdev, p_blkno, osb->sb->s_blocksize); /* block not cached. */ if (!bh) diff --git a/fs/splice.c b/fs/splice.c index 90d464241f15..4d6df083e0c0 100644 --- a/fs/splice.c +++ b/fs/splice.c @@ -45,7 +45,7 @@ * here if set to avoid blocking other users of this pipe if splice is * being done on it. */ -static noinline void noinline pipe_clear_nowait(struct file *file) +static noinline void pipe_clear_nowait(struct file *file) { fmode_t fmode = READ_ONCE(file->f_mode); diff --git a/fs/xattr.c b/fs/xattr.c index 02bee149ad96..fabb2a04501e 100644 --- a/fs/xattr.c +++ b/fs/xattr.c @@ -703,7 +703,7 @@ static int path_setxattrat(int dfd, const char __user *pathname, return error; filename = getname_maybe_null(pathname, at_flags); - if (!filename) { + if (!filename && dfd >= 0) { CLASS(fd, f)(dfd); if (fd_empty(f)) error = -EBADF; @@ -847,7 +847,7 @@ static ssize_t path_getxattrat(int dfd, const char __user *pathname, return error; filename = getname_maybe_null(pathname, at_flags); - if (!filename) { + if (!filename && dfd >= 0) { CLASS(fd, f)(dfd); if (fd_empty(f)) return -EBADF; diff --git a/fs/xfs/xfs_zone_gc.c b/fs/xfs/xfs_zone_gc.c index 8c541ca71872..81c94dd1d596 100644 --- a/fs/xfs/xfs_zone_gc.c +++ b/fs/xfs/xfs_zone_gc.c @@ -170,7 +170,8 @@ bool xfs_zoned_need_gc( struct xfs_mount *mp) { - s64 available, free; + s64 available, free, threshold; + s32 remainder; if (!xfs_group_marked(mp, XG_TYPE_RTG, XFS_RTG_RECLAIMABLE)) return false; @@ -183,7 +184,12 @@ xfs_zoned_need_gc( return true; free = xfs_estimate_freecounter(mp, XC_FREE_RTEXTENTS); - if (available < mult_frac(free, mp->m_zonegc_low_space, 100)) + + threshold = div_s64_rem(free, 100, &remainder); + threshold = threshold * mp->m_zonegc_low_space + + remainder * div_s64(mp->m_zonegc_low_space, 100); + + if (available < threshold) return true; return false; |