xfs_rmap_btree.c | Explore in Territory

// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (c) 2014 Red Hat, Inc.
 * All Rights Reserved.
 */
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_shared.h"
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_mount.h"
#include "xfs_trans.h"
#include "xfs_alloc.h"
#include "xfs_btree.h"
#include "xfs_btree_staging.h"
#include "xfs_rmap.h"
#include "xfs_rmap_btree.h"
#include "xfs_health.h"
#include "xfs_trace.h"
#include "xfs_error.h"
#include "xfs_extent_busy.h"
#include "xfs_ag.h"
#include "xfs_ag_resv.h"
#include "xfs_buf_mem.h"
#include "xfs_btree_mem.h"

static struct kmem_cache	*xfs_rmapbt_cur_cache;

/*
 * Reverse map btree.
 *
 * This is a per-ag tree used to track the owner(s) of a given extent. With
 * reflink it is possible for there to be multiple owners, which is a departure
 * from classic XFS. Owner records for data extents are inserted when the
 * extent is mapped and removed when an extent is unmapped.  Owner records for
 * all other block types (i.e. metadata) are inserted when an extent is
 * allocated and removed when an extent is freed. There can only be one owner
 * of a metadata extent, usually an inode or some other metadata structure like
 * an AG btree.
 *
 * The rmap btree is part of the free space management, so blocks for the tree
 * are sourced from the agfl. Hence we need transaction reservation support for
 * this tree so that the freelist is always large enough. This also impacts on
 * the minimum space we need to leave free in the AG.
 *
 * The tree is ordered by [ag block, owner, offset]. This is a large key size,
 * but it is the only way to enforce unique keys when a block can be owned by
 * multiple files at any offset. There's no need to order/search by extent
 * size for online updating/management of the tree. It is intended that most
 * reverse lookups will be to find the owner(s) of a particular block, or to
 * try to recover tree and file data from corrupt primary metadata.
 */

static struct xfs_btree_cur *
xfs_rmapbt_dup_cursor(
	struct xfs_btree_cur	*cur)
{ … }

STATIC void
xfs_rmapbt_set_root(
	struct xfs_btree_cur		*cur,
	const union xfs_btree_ptr	*ptr,
	int				inc)
{ … }

STATIC int
xfs_rmapbt_alloc_block(
	struct xfs_btree_cur		*cur,
	const union xfs_btree_ptr	*start,
	union xfs_btree_ptr		*new,
	int				*stat)
{ … }

STATIC int
xfs_rmapbt_free_block(
	struct xfs_btree_cur	*cur,
	struct xfs_buf		*bp)
{ … }

STATIC int
xfs_rmapbt_get_minrecs(
	struct xfs_btree_cur	*cur,
	int			level)
{ … }

STATIC int
xfs_rmapbt_get_maxrecs(
	struct xfs_btree_cur	*cur,
	int			level)
{ … }

/*
 * Convert the ondisk record's offset field into the ondisk key's offset field.
 * Fork and bmbt are significant parts of the rmap record key, but written
 * status is merely a record attribute.
 */
static inline __be64 ondisk_rec_offset_to_key(const union xfs_btree_rec *rec)
{ … }

STATIC void
xfs_rmapbt_init_key_from_rec(
	union xfs_btree_key		*key,
	const union xfs_btree_rec	*rec)
{ … }

/*
 * The high key for a reverse mapping record can be computed by shifting
 * the startblock and offset to the highest value that would still map
 * to that record.  In practice this means that we add blockcount-1 to
 * the startblock for all records, and if the record is for a data/attr
 * fork mapping, we add blockcount-1 to the offset too.
 */
STATIC void
xfs_rmapbt_init_high_key_from_rec(
	union xfs_btree_key		*key,
	const union xfs_btree_rec	*rec)
{ … }

STATIC void
xfs_rmapbt_init_rec_from_cur(
	struct xfs_btree_cur	*cur,
	union xfs_btree_rec	*rec)
{ … }

STATIC void
xfs_rmapbt_init_ptr_from_cur(
	struct xfs_btree_cur	*cur,
	union xfs_btree_ptr	*ptr)
{ … }

/*
 * Mask the appropriate parts of the ondisk key field for a key comparison.
 * Fork and bmbt are significant parts of the rmap record key, but written
 * status is merely a record attribute.
 */
static inline uint64_t offset_keymask(uint64_t offset)
{ … }

STATIC int64_t
xfs_rmapbt_key_diff(
	struct xfs_btree_cur		*cur,
	const union xfs_btree_key	*key)
{ … }

STATIC int64_t
xfs_rmapbt_diff_two_keys(
	struct xfs_btree_cur		*cur,
	const union xfs_btree_key	*k1,
	const union xfs_btree_key	*k2,
	const union xfs_btree_key	*mask)
{ … }

static xfs_failaddr_t
xfs_rmapbt_verify(
	struct xfs_buf		*bp)
{ … }

static void
xfs_rmapbt_read_verify(
	struct xfs_buf	*bp)
{ … }

static void
xfs_rmapbt_write_verify(
	struct xfs_buf	*bp)
{ … }

const struct xfs_buf_ops xfs_rmapbt_buf_ops = …;

STATIC int
xfs_rmapbt_keys_inorder(
	struct xfs_btree_cur		*cur,
	const union xfs_btree_key	*k1,
	const union xfs_btree_key	*k2)
{ … }

STATIC int
xfs_rmapbt_recs_inorder(
	struct xfs_btree_cur		*cur,
	const union xfs_btree_rec	*r1,
	const union xfs_btree_rec	*r2)
{ … }

STATIC enum xbtree_key_contig
xfs_rmapbt_keys_contiguous(
	struct xfs_btree_cur		*cur,
	const union xfs_btree_key	*key1,
	const union xfs_btree_key	*key2,
	const union xfs_btree_key	*mask)
{ … }

const struct xfs_btree_ops xfs_rmapbt_ops = …;

/*
 * Create a new reverse mapping btree cursor.
 *
 * For staging cursors tp and agbp are NULL.
 */
struct xfs_btree_cur *
xfs_rmapbt_init_cursor(
	struct xfs_mount	*mp,
	struct xfs_trans	*tp,
	struct xfs_buf		*agbp,
	struct xfs_perag	*pag)
{ … }

#ifdef CONFIG_XFS_BTREE_IN_MEM
static inline unsigned int
xfs_rmapbt_mem_block_maxrecs(
	unsigned int		blocklen,
	bool			leaf)
{ … }

/*
 * Validate an in-memory rmap btree block.  Callers are allowed to generate an
 * in-memory btree even if the ondisk feature is not enabled.
 */
static xfs_failaddr_t
xfs_rmapbt_mem_verify(
	struct xfs_buf		*bp)
{ … }

static void
xfs_rmapbt_mem_rw_verify(
	struct xfs_buf	*bp)
{ … }

/* skip crc checks on in-memory btrees to save time */
static const struct xfs_buf_ops xfs_rmapbt_mem_buf_ops = …;

const struct xfs_btree_ops xfs_rmapbt_mem_ops = …;

/* Create a cursor for an in-memory btree. */
struct xfs_btree_cur *
xfs_rmapbt_mem_cursor(
	struct xfs_perag	*pag,
	struct xfs_trans	*tp,
	struct xfbtree		*xfbt)
{ … }

/* Create an in-memory rmap btree. */
int
xfs_rmapbt_mem_init(
	struct xfs_mount	*mp,
	struct xfbtree		*xfbt,
	struct xfs_buftarg	*btp,
	xfs_agnumber_t		agno)
{ … }

/* Compute the max possible height for reverse mapping btrees in memory. */
static unsigned int
xfs_rmapbt_mem_maxlevels(void)
{ … }
#else
#define xfs_rmapbt_mem_maxlevels …
#endif /* CONFIG_XFS_BTREE_IN_MEM */

/*
 * Install a new reverse mapping btree root.  Caller is responsible for
 * invalidating and freeing the old btree blocks.
 */
void
xfs_rmapbt_commit_staged_btree(
	struct xfs_btree_cur	*cur,
	struct xfs_trans	*tp,
	struct xfs_buf		*agbp)
{ … }

/* Calculate number of records in a reverse mapping btree block. */
static inline unsigned int
xfs_rmapbt_block_maxrecs(
	unsigned int		blocklen,
	bool			leaf)
{ … }

/*
 * Calculate number of records in an rmap btree block.
 */
int
xfs_rmapbt_maxrecs(
	int			blocklen,
	int			leaf)
{ … }

/* Compute the max possible height for reverse mapping btrees. */
unsigned int
xfs_rmapbt_maxlevels_ondisk(void)
{ … }

/* Compute the maximum height of an rmap btree. */
void
xfs_rmapbt_compute_maxlevels(
	struct xfs_mount		*mp)
{ … }

/* Calculate the refcount btree size for some records. */
xfs_extlen_t
xfs_rmapbt_calc_size(
	struct xfs_mount	*mp,
	unsigned long long	len)
{ … }

/*
 * Calculate the maximum refcount btree size.
 */
xfs_extlen_t
xfs_rmapbt_max_size(
	struct xfs_mount	*mp,
	xfs_agblock_t		agblocks)
{ … }

/*
 * Figure out how many blocks to reserve and how many are used by this btree.
 */
int
xfs_rmapbt_calc_reserves(
	struct xfs_mount	*mp,
	struct xfs_trans	*tp,
	struct xfs_perag	*pag,
	xfs_extlen_t		*ask,
	xfs_extlen_t		*used)
{ … }

int __init
xfs_rmapbt_init_cur_cache(void)
{ … }

void
xfs_rmapbt_destroy_cur_cache(void)
{ … }
linux/fs/xfs/libxfs/xfs_rmap_btree.c
