#include <asm/byteorder.h>
#include <linux/kernel.h>
#include <linux/uuid.h>
+#include "vstructs.h"
+
+#define BITMASK(name, type, field, offset, end) \
+static const unsigned name##_OFFSET = offset; \
+static const unsigned name##_BITS = (end - offset); \
+ \
+static inline __u64 name(const type *k) \
+{ \
+ return (k->field >> offset) & ~(~0ULL << (end - offset)); \
+} \
+ \
+static inline void SET_##name(type *k, __u64 v) \
+{ \
+ k->field &= ~(~(~0ULL << (end - offset)) << offset); \
+ k->field |= (v & ~(~0ULL << (end - offset))) << offset; \
+}
#define LE_BITMASK(_bits, name, type, field, offset, end) \
static const unsigned name##_OFFSET = offset; \
}
#define POS_MIN SPOS(0, 0, 0)
-#define POS_MAX SPOS(KEY_INODE_MAX, KEY_OFFSET_MAX, KEY_SNAPSHOT_MAX)
+#define POS_MAX SPOS(KEY_INODE_MAX, KEY_OFFSET_MAX, 0)
+#define SPOS_MAX SPOS(KEY_INODE_MAX, KEY_OFFSET_MAX, KEY_SNAPSHOT_MAX)
#define POS(_inode, _offset) SPOS(_inode, _offset, 0)
/* Empty placeholder struct, for container_of() */
*/
#define BCH_BKEY_TYPES() \
x(deleted, 0) \
- x(discard, 1) \
+ x(whiteout, 1) \
x(error, 2) \
x(cookie, 3) \
x(hash_whiteout, 4) \
x(inline_data, 17) \
x(btree_ptr_v2, 18) \
x(indirect_inline_data, 19) \
- x(alloc_v2, 20)
+ 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)
enum bch_bkey_type {
#define x(name, nr) KEY_TYPE_##name = nr,
struct bch_val v;
};
-struct bch_discard {
+struct bch_whiteout {
struct bch_val v;
};
struct bch_val v;
};
+struct bch_set {
+ struct bch_val v;
+};
+
/* Extents */
/*
__u8 fields[0];
} __attribute__((packed, aligned(8)));
+struct bch_inode_v2 {
+ struct bch_val v;
+
+ __le64 bi_journal_seq;
+ __le64 bi_hash_seed;
+ __le64 bi_flags;
+ __le16 bi_mode;
+ __u8 fields[0];
+} __attribute__((packed, aligned(8)));
+
struct bch_inode_generation {
struct bch_val v;
__le32 pad;
} __attribute__((packed, aligned(8)));
+/*
+ * bi_subvol and bi_parent_subvol are only set for subvolume roots:
+ */
+
#define BCH_INODE_FIELDS() \
x(bi_atime, 96) \
x(bi_ctime, 96) \
x(bi_erasure_code, 16) \
x(bi_fields_set, 16) \
x(bi_dir, 64) \
- x(bi_dir_offset, 64)
+ x(bi_dir_offset, 64) \
+ x(bi_subvol, 32) \
+ x(bi_parent_subvol, 32)
/* subset of BCH_INODE_FIELDS */
#define BCH_INODE_OPTS() \
LE32_BITMASK(INODE_NR_FIELDS, struct bch_inode, bi_flags, 24, 31);
LE32_BITMASK(INODE_NEW_VARINT, struct bch_inode, bi_flags, 31, 32);
+LE64_BITMASK(INODEv2_STR_HASH, struct bch_inode_v2, bi_flags, 20, 24);
+LE64_BITMASK(INODEv2_NR_FIELDS, struct bch_inode_v2, bi_flags, 24, 31);
+
/* Dirents */
/*
struct bch_val v;
/* Target inode number: */
+ union {
__le64 d_inum;
+ struct { /* DT_SUBVOL */
+ __le32 d_child_subvol;
+ __le32 d_parent_subvol;
+ };
+ };
/*
* Copy of mode bits 12-15 from the target inode - so userspace can get
__u8 d_name[];
} __attribute__((packed, aligned(8)));
+#define DT_SUBVOL 16
+#define BCH_DT_MAX 17
+
#define BCH_NAME_MAX (U8_MAX * sizeof(u64) - \
sizeof(struct bkey) - \
offsetof(struct bch_dirent, d_name))
#define BCH_ALLOC_FIELDS_V2() \
x(read_time, 64) \
x(write_time, 64) \
- x(dirty_sectors, 16) \
- x(cached_sectors, 16) \
+ x(dirty_sectors, 32) \
+ x(cached_sectors, 32) \
x(stripe, 32) \
x(stripe_redundancy, 8)
+struct bch_alloc_v3 {
+ struct bch_val v;
+ __le64 journal_seq;
+ __le32 flags;
+ __u8 nr_fields;
+ __u8 gen;
+ __u8 oldest_gen;
+ __u8 data_type;
+ __u8 data[];
+} __attribute__((packed, aligned(8)));
+
+struct bch_alloc_v4 {
+ struct bch_val v;
+ __u64 journal_seq;
+ __u32 flags;
+ __u8 gen;
+ __u8 oldest_gen;
+ __u8 data_type;
+ __u8 stripe_redundancy;
+ __u32 dirty_sectors;
+ __u32 cached_sectors;
+ __u64 io_time[2];
+ __u32 stripe;
+ __u32 nr_external_backpointers;
+ struct bpos backpointers[0];
+} __attribute__((packed, aligned(8)));
+
+LE32_BITMASK(BCH_ALLOC_V3_NEED_DISCARD,struct bch_alloc_v3, flags, 0, 1)
+LE32_BITMASK(BCH_ALLOC_V3_NEED_INC_GEN,struct bch_alloc_v3, flags, 1, 2)
+
+BITMASK(BCH_ALLOC_V4_NEED_DISCARD, struct bch_alloc_v4, flags, 0, 1)
+BITMASK(BCH_ALLOC_V4_NEED_INC_GEN, struct bch_alloc_v4, flags, 1, 2)
+BITMASK(BCH_ALLOC_V4_BACKPOINTERS_START,struct bch_alloc_v4, flags, 2, 8)
+BITMASK(BCH_ALLOC_V4_NR_BACKPOINTERS, struct bch_alloc_v4, flags, 8, 14)
+
enum {
#define x(name, _bits) BCH_ALLOC_FIELD_V1_##name,
BCH_ALLOC_FIELDS_V1()
#undef x
- BCH_ALLOC_FIELD_NR
};
/* Quotas: */
struct bch_reflink_p {
struct bch_val v;
__le64 idx;
-
- __le32 reservation_generation;
- __u8 nr_replicas;
- __u8 pad[3];
-};
+ /*
+ * A reflink pointer might point to an indirect extent which is then
+ * later split (by copygc or rebalance). If we only pointed to part of
+ * the original indirect extent, and then one of the fragments is
+ * outside the range we point to, we'd leak a refcount: so when creating
+ * reflink pointers, we need to store pad values to remember the full
+ * range we were taking a reference on.
+ */
+ __le32 front_pad;
+ __le32 back_pad;
+} __attribute__((packed, aligned(8)));
struct bch_reflink_v {
struct bch_val v;
__le64 refcount;
union bch_extent_entry start[0];
__u64 _data[0];
-};
+} __attribute__((packed, aligned(8)));
struct bch_indirect_inline_data {
struct bch_val v;
u8 data[0];
};
+/* Subvolumes: */
+
+#define SUBVOL_POS_MIN POS(0, 1)
+#define SUBVOL_POS_MAX POS(0, S32_MAX)
+#define BCACHEFS_ROOT_SUBVOL 1
+
+struct bch_subvolume {
+ struct bch_val v;
+ __le32 flags;
+ __le32 snapshot;
+ __le64 inode;
+};
+
+LE32_BITMASK(BCH_SUBVOLUME_RO, struct bch_subvolume, flags, 0, 1)
+/*
+ * We need to know whether a subvolume is a snapshot so we can know whether we
+ * can delete it (or whether it should just be rm -rf'd)
+ */
+LE32_BITMASK(BCH_SUBVOLUME_SNAP, struct bch_subvolume, flags, 1, 2)
+LE32_BITMASK(BCH_SUBVOLUME_UNLINKED, struct bch_subvolume, flags, 2, 3)
+
+/* Snapshots */
+
+struct bch_snapshot {
+ struct bch_val v;
+ __le32 flags;
+ __le32 parent;
+ __le32 children[2];
+ __le32 subvol;
+ __le32 pad;
+};
+
+LE32_BITMASK(BCH_SNAPSHOT_DELETED, struct bch_snapshot, flags, 0, 1)
+
+/* True if a subvolume points to this snapshot node: */
+LE32_BITMASK(BCH_SNAPSHOT_SUBVOL, struct bch_snapshot, flags, 1, 2)
+
+/* LRU btree: */
+
+struct bch_lru {
+ struct bch_val v;
+ __le64 idx;
+} __attribute__((packed, aligned(8)));
+
+#define LRU_ID_STRIPES (1U << 16)
+
/* Optional/variable size superblock sections: */
struct bch_sb_field {
__le32 type;
};
-#define BCH_SB_FIELDS() \
- x(journal, 0) \
- x(members, 1) \
- x(crypt, 2) \
- x(replicas_v0, 3) \
- x(quota, 4) \
- x(disk_groups, 5) \
- x(clean, 6) \
- x(replicas, 7) \
- x(journal_seq_blacklist, 8)
+#define BCH_SB_FIELDS() \
+ x(journal, 0) \
+ x(members, 1) \
+ x(crypt, 2) \
+ x(replicas_v0, 3) \
+ x(quota, 4) \
+ x(disk_groups, 5) \
+ x(clean, 6) \
+ x(replicas, 7) \
+ x(journal_seq_blacklist, 8) \
+ x(journal_v2, 9)
enum bch_sb_field_type {
#define x(f, nr) BCH_SB_FIELD_##f = nr,
BCH_SB_FIELD_NR
};
+/*
+ * Most superblock fields are replicated in all device's superblocks - a few are
+ * not:
+ */
+#define BCH_SINGLE_DEVICE_SB_FIELDS \
+ ((1U << BCH_SB_FIELD_journal)| \
+ (1U << BCH_SB_FIELD_journal_v2))
+
/* BCH_SB_FIELD_journal: */
struct bch_sb_field_journal {
__le64 buckets[0];
};
+struct bch_sb_field_journal_v2 {
+ struct bch_sb_field field;
+
+ struct bch_sb_field_journal_v2_entry {
+ __le64 start;
+ __le64 nr;
+ } d[0];
+};
+
/* BCH_SB_FIELD_members: */
#define BCH_MIN_NR_NBUCKETS (1 << 6)
};
LE64_BITMASK(BCH_MEMBER_STATE, struct bch_member, flags[0], 0, 4)
-/* 4-10 unused, was TIER, HAS_(META)DATA */
-LE64_BITMASK(BCH_MEMBER_REPLACEMENT, struct bch_member, flags[0], 10, 14)
+/* 4-14 unused, was TIER, HAS_(META)DATA, REPLACEMENT */
LE64_BITMASK(BCH_MEMBER_DISCARD, struct bch_member, flags[0], 14, 15)
LE64_BITMASK(BCH_MEMBER_DATA_ALLOWED, struct bch_member, flags[0], 15, 20)
LE64_BITMASK(BCH_MEMBER_GROUP, struct bch_member, flags[0], 20, 28)
LE64_BITMASK(BCH_MEMBER_DURABILITY, struct bch_member, flags[0], 28, 30)
-
-#define BCH_TIER_MAX 4U
+LE64_BITMASK(BCH_MEMBER_FREESPACE_INITIALIZED,
+ struct bch_member, flags[0], 30, 31)
#if 0
LE64_BITMASK(BCH_MEMBER_NR_READ_ERRORS, struct bch_member, flags[1], 0, 20);
BCH_MEMBER_STATE_NR
};
-#define BCH_CACHE_REPLACEMENT_POLICIES() \
- x(lru, 0) \
- x(fifo, 1) \
- x(random, 2)
-
-enum bch_cache_replacement_policies {
-#define x(t, n) BCH_CACHE_REPLACEMENT_##t = n,
- BCH_CACHE_REPLACEMENT_POLICIES()
-#undef x
- BCH_CACHE_REPLACEMENT_NR
-};
-
struct bch_sb_field_members {
struct bch_sb_field field;
struct bch_member members[0];
#define BCH_JSET_VERSION_OLD 2
#define BCH_BSET_VERSION_OLD 3
+#define BCH_METADATA_VERSIONS() \
+ x(bkey_renumber, 10) \
+ x(inode_btree_change, 11) \
+ x(snapshot, 12) \
+ x(inode_backpointers, 13) \
+ x(btree_ptr_sectors_written, 14) \
+ x(snapshot_2, 15) \
+ x(reflink_p_fix, 16) \
+ x(subvol_dirent, 17) \
+ x(inode_v2, 18) \
+ x(freespace, 19) \
+ x(alloc_v4, 20)
+
enum bcachefs_metadata_version {
- bcachefs_metadata_version_min = 9,
- bcachefs_metadata_version_new_versioning = 10,
- bcachefs_metadata_version_bkey_renumber = 10,
- bcachefs_metadata_version_inode_btree_change = 11,
- bcachefs_metadata_version_snapshot = 12,
- bcachefs_metadata_version_inode_backpointers = 13,
- bcachefs_metadata_version_max = 14,
+ bcachefs_metadata_version_min = 9,
+#define x(t, n) bcachefs_metadata_version_##t = n,
+ BCH_METADATA_VERSIONS()
+#undef x
+ bcachefs_metadata_version_max
};
#define bcachefs_metadata_version_current (bcachefs_metadata_version_max - 1)
LE64_BITMASK(BCH_SB_ERASURE_CODE, struct bch_sb, flags[3], 0, 16);
LE64_BITMASK(BCH_SB_METADATA_TARGET, struct bch_sb, flags[3], 16, 28);
LE64_BITMASK(BCH_SB_SHARD_INUMS, struct bch_sb, flags[3], 28, 29);
+LE64_BITMASK(BCH_SB_INODES_USE_KEY_CACHE,struct bch_sb, flags[3], 29, 30);
+LE64_BITMASK(BCH_SB_JOURNAL_FLUSH_DELAY,struct bch_sb, flags[3], 30, 62);
+LE64_BITMASK(BCH_SB_JOURNAL_FLUSH_DISABLED,struct bch_sb, flags[3], 62, 63);
+LE64_BITMASK(BCH_SB_JOURNAL_RECLAIM_DELAY,struct bch_sb, flags[4], 0, 32);
+LE64_BITMASK(BCH_SB_JOURNAL_TRANSACTION_NAMES,struct bch_sb, flags[4], 32, 33);
/*
* Features:
* journal_seq_blacklist_v3: gates BCH_SB_FIELD_journal_seq_blacklist
* reflink: gates KEY_TYPE_reflink
* inline_data: gates KEY_TYPE_inline_data
- * new_siphash: gates BCH_STR_HASH_SIPHASH
+ * new_siphash: gates BCH_STR_HASH_siphash
* new_extent_overwrite: gates BTREE_NODE_NEW_EXTENT_OVERWRITE
*/
#define BCH_SB_FEATURES() \
BCH_ON_ERROR_NR
};
+#define BCH_STR_HASH_TYPES() \
+ x(crc32c, 0) \
+ x(crc64, 1) \
+ x(siphash_old, 2) \
+ x(siphash, 3)
+
enum bch_str_hash_type {
- BCH_STR_HASH_CRC32C = 0,
- BCH_STR_HASH_CRC64 = 1,
- BCH_STR_HASH_SIPHASH_OLD = 2,
- BCH_STR_HASH_SIPHASH = 3,
- BCH_STR_HASH_NR = 4,
+#define x(t, n) BCH_STR_HASH_##t = n,
+ BCH_STR_HASH_TYPES()
+#undef x
+ BCH_STR_HASH_NR
};
#define BCH_STR_HASH_OPTS() \
BCH_STR_HASH_OPT_NR
};
+#define BCH_CSUM_TYPES() \
+ x(none, 0) \
+ x(crc32c_nonzero, 1) \
+ x(crc64_nonzero, 2) \
+ x(chacha20_poly1305_80, 3) \
+ x(chacha20_poly1305_128, 4) \
+ x(crc32c, 5) \
+ x(crc64, 6) \
+ x(xxhash, 7)
+
enum bch_csum_type {
- BCH_CSUM_NONE = 0,
- BCH_CSUM_CRC32C_NONZERO = 1,
- BCH_CSUM_CRC64_NONZERO = 2,
- BCH_CSUM_CHACHA20_POLY1305_80 = 3,
- BCH_CSUM_CHACHA20_POLY1305_128 = 4,
- BCH_CSUM_CRC32C = 5,
- BCH_CSUM_CRC64 = 6,
- BCH_CSUM_XXHASH = 7,
- BCH_CSUM_NR = 8,
+#define x(t, n) BCH_CSUM_##t = n,
+ BCH_CSUM_TYPES()
+#undef x
+ BCH_CSUM_NR
};
static const unsigned bch_crc_bytes[] = {
- [BCH_CSUM_NONE] = 0,
- [BCH_CSUM_CRC32C_NONZERO] = 4,
- [BCH_CSUM_CRC32C] = 4,
- [BCH_CSUM_CRC64_NONZERO] = 8,
- [BCH_CSUM_CRC64] = 8,
- [BCH_CSUM_XXHASH] = 8,
- [BCH_CSUM_CHACHA20_POLY1305_80] = 10,
- [BCH_CSUM_CHACHA20_POLY1305_128] = 16,
+ [BCH_CSUM_none] = 0,
+ [BCH_CSUM_crc32c_nonzero] = 4,
+ [BCH_CSUM_crc32c] = 4,
+ [BCH_CSUM_crc64_nonzero] = 8,
+ [BCH_CSUM_crc64] = 8,
+ [BCH_CSUM_xxhash] = 8,
+ [BCH_CSUM_chacha20_poly1305_80] = 10,
+ [BCH_CSUM_chacha20_poly1305_128] = 16,
};
static inline _Bool bch2_csum_type_is_encryption(enum bch_csum_type type)
{
switch (type) {
- case BCH_CSUM_CHACHA20_POLY1305_80:
- case BCH_CSUM_CHACHA20_POLY1305_128:
+ case BCH_CSUM_chacha20_poly1305_80:
+ case BCH_CSUM_chacha20_poly1305_128:
return true;
default:
return false;
x(usage, 5) \
x(data_usage, 6) \
x(clock, 7) \
- x(dev_usage, 8)
+ x(dev_usage, 8) \
+ x(log, 9)
enum {
#define x(f, nr) BCH_JSET_ENTRY_##f = nr,
__le64 end;
};
+#define BCH_FS_USAGE_TYPES() \
+ x(reserved, 0) \
+ x(inodes, 1) \
+ x(key_version, 2)
+
enum {
- FS_USAGE_RESERVED = 0,
- FS_USAGE_INODES = 1,
- FS_USAGE_KEY_VERSION = 2,
- FS_USAGE_NR = 3
+#define x(f, nr) BCH_FS_USAGE_##f = nr,
+ BCH_FS_USAGE_TYPES()
+#undef x
+ BCH_FS_USAGE_NR
};
struct jset_entry_usage {
struct jset_entry_dev_usage_type d[];
} __attribute__((packed));
+static inline unsigned jset_entry_dev_usage_nr_types(struct jset_entry_dev_usage *u)
+{
+ return (vstruct_bytes(&u->entry) - sizeof(struct jset_entry_dev_usage)) /
+ sizeof(struct jset_entry_dev_usage_type);
+}
+
+struct jset_entry_log {
+ struct jset_entry entry;
+ u8 d[];
+} __attribute__((packed));
+
/*
* On disk format for a journal entry:
* seq is monotonically increasing; every journal entry has its own unique
x(alloc, 4) \
x(quotas, 5) \
x(stripes, 6) \
- x(reflink, 7)
+ x(reflink, 7) \
+ x(subvolumes, 8) \
+ x(snapshots, 9) \
+ x(lru, 10) \
+ x(freespace, 11) \
+ x(need_discard, 12) \
+ x(backpointers, 13)
enum btree_id {
#define x(kwd, val) BTREE_ID_##kwd = val,
LE32_BITMASK(BSET_SEPARATE_WHITEOUTS,
struct bset, flags, 5, 6);
+/* Sector offset within the btree node: */
+LE32_BITMASK(BSET_OFFSET, struct bset, flags, 16, 32);
+
struct btree_node {
struct bch_csum csum;
__le64 magic;