aboutsummaryrefslogtreecommitdiff
path: root/fs/reiserfs/reiserfs.h
diff options
context:
space:
mode:
Diffstat (limited to 'fs/reiserfs/reiserfs.h')
-rw-r--r--fs/reiserfs/reiserfs.h2327
1 files changed, 2327 insertions, 0 deletions
diff --git a/fs/reiserfs/reiserfs.h b/fs/reiserfs/reiserfs.h
new file mode 100644
index 00000000000..b3865c84f54
--- /dev/null
+++ b/fs/reiserfs/reiserfs.h
@@ -0,0 +1,2327 @@
+/*
+ * Copyright 1996, 1997, 1998 Hans Reiser, see reiserfs/README for licensing and copyright details
+ */
+
+#include <linux/reiserfs_fs.h>
+
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/workqueue.h>
+#include <asm/unaligned.h>
+#include <linux/bitops.h>
+#include <linux/proc_fs.h>
+#include <linux/buffer_head.h>
+#include <linux/reiserfs_fs_i.h>
+#include <linux/reiserfs_fs_sb.h>
+
+/* the 32 bit compat definitions with int argument */
+#define REISERFS_IOC32_UNPACK _IOW(0xCD, 1, int)
+#define REISERFS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
+#define REISERFS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
+#define REISERFS_IOC32_GETVERSION FS_IOC32_GETVERSION
+#define REISERFS_IOC32_SETVERSION FS_IOC32_SETVERSION
+
+/*
+ * Locking primitives. The write lock is a per superblock
+ * special mutex that has properties close to the Big Kernel Lock
+ * which was used in the previous locking scheme.
+ */
+void reiserfs_write_lock(struct super_block *s);
+void reiserfs_write_unlock(struct super_block *s);
+int reiserfs_write_lock_once(struct super_block *s);
+void reiserfs_write_unlock_once(struct super_block *s, int lock_depth);
+
+#ifdef CONFIG_REISERFS_CHECK
+void reiserfs_lock_check_recursive(struct super_block *s);
+#else
+static inline void reiserfs_lock_check_recursive(struct super_block *s) { }
+#endif
+
+/*
+ * Several mutexes depend on the write lock.
+ * However sometimes we want to relax the write lock while we hold
+ * these mutexes, according to the release/reacquire on schedule()
+ * properties of the Bkl that were used.
+ * Reiserfs performances and locking were based on this scheme.
+ * Now that the write lock is a mutex and not the bkl anymore, doing so
+ * may result in a deadlock:
+ *
+ * A acquire write_lock
+ * A acquire j_commit_mutex
+ * A release write_lock and wait for something
+ * B acquire write_lock
+ * B can't acquire j_commit_mutex and sleep
+ * A can't acquire write lock anymore
+ * deadlock
+ *
+ * What we do here is avoiding such deadlock by playing the same game
+ * than the Bkl: if we can't acquire a mutex that depends on the write lock,
+ * we release the write lock, wait a bit and then retry.
+ *
+ * The mutexes concerned by this hack are:
+ * - The commit mutex of a journal list
+ * - The flush mutex
+ * - The journal lock
+ * - The inode mutex
+ */
+static inline void reiserfs_mutex_lock_safe(struct mutex *m,
+ struct super_block *s)
+{
+ reiserfs_lock_check_recursive(s);
+ reiserfs_write_unlock(s);
+ mutex_lock(m);
+ reiserfs_write_lock(s);
+}
+
+static inline void
+reiserfs_mutex_lock_nested_safe(struct mutex *m, unsigned int subclass,
+ struct super_block *s)
+{
+ reiserfs_lock_check_recursive(s);
+ reiserfs_write_unlock(s);
+ mutex_lock_nested(m, subclass);
+ reiserfs_write_lock(s);
+}
+
+static inline void
+reiserfs_down_read_safe(struct rw_semaphore *sem, struct super_block *s)
+{
+ reiserfs_lock_check_recursive(s);
+ reiserfs_write_unlock(s);
+ down_read(sem);
+ reiserfs_write_lock(s);
+}
+
+/*
+ * When we schedule, we usually want to also release the write lock,
+ * according to the previous bkl based locking scheme of reiserfs.
+ */
+static inline void reiserfs_cond_resched(struct super_block *s)
+{
+ if (need_resched()) {
+ reiserfs_write_unlock(s);
+ schedule();
+ reiserfs_write_lock(s);
+ }
+}
+
+struct fid;
+
+/* in reading the #defines, it may help to understand that they employ
+ the following abbreviations:
+
+ B = Buffer
+ I = Item header
+ H = Height within the tree (should be changed to LEV)
+ N = Number of the item in the node
+ STAT = stat data
+ DEH = Directory Entry Header
+ EC = Entry Count
+ E = Entry number
+ UL = Unsigned Long
+ BLKH = BLocK Header
+ UNFM = UNForMatted node
+ DC = Disk Child
+ P = Path
+
+ These #defines are named by concatenating these abbreviations,
+ where first comes the arguments, and last comes the return value,
+ of the macro.
+
+*/
+
+#define USE_INODE_GENERATION_COUNTER
+
+#define REISERFS_PREALLOCATE
+#define DISPLACE_NEW_PACKING_LOCALITIES
+#define PREALLOCATION_SIZE 9
+
+/* n must be power of 2 */
+#define _ROUND_UP(x,n) (((x)+(n)-1u) & ~((n)-1u))
+
+// to be ok for alpha and others we have to align structures to 8 byte
+// boundary.
+// FIXME: do not change 4 by anything else: there is code which relies on that
+#define ROUND_UP(x) _ROUND_UP(x,8LL)
+
+/* debug levels. Right now, CONFIG_REISERFS_CHECK means print all debug
+** messages.
+*/
+#define REISERFS_DEBUG_CODE 5 /* extra messages to help find/debug errors */
+
+void __reiserfs_warning(struct super_block *s, const char *id,
+ const char *func, const char *fmt, ...);
+#define reiserfs_warning(s, id, fmt, args...) \
+ __reiserfs_warning(s, id, __func__, fmt, ##args)
+/* assertions handling */
+
+/** always check a condition and panic if it's false. */
+#define __RASSERT(cond, scond, format, args...) \
+do { \
+ if (!(cond)) \
+ reiserfs_panic(NULL, "assertion failure", "(" #cond ") at " \
+ __FILE__ ":%i:%s: " format "\n", \
+ in_interrupt() ? -1 : task_pid_nr(current), \
+ __LINE__, __func__ , ##args); \
+} while (0)
+
+#define RASSERT(cond, format, args...) __RASSERT(cond, #cond, format, ##args)
+
+#if defined( CONFIG_REISERFS_CHECK )
+#define RFALSE(cond, format, args...) __RASSERT(!(cond), "!(" #cond ")", format, ##args)
+#else
+#define RFALSE( cond, format, args... ) do {;} while( 0 )
+#endif
+
+#define CONSTF __attribute_const__
+/*
+ * Disk Data Structures
+ */
+
+/***************************************************************************/
+/* SUPER BLOCK */
+/***************************************************************************/
+
+/*
+ * Structure of super block on disk, a version of which in RAM is often accessed as REISERFS_SB(s)->s_rs
+ * the version in RAM is part of a larger structure containing fields never written to disk.
+ */
+#define UNSET_HASH 0 // read_super will guess about, what hash names
+ // in directories were sorted with
+#define TEA_HASH 1
+#define YURA_HASH 2
+#define R5_HASH 3
+#define DEFAULT_HASH R5_HASH
+
+struct journal_params {
+ __le32 jp_journal_1st_block; /* where does journal start from on its
+ * device */
+ __le32 jp_journal_dev; /* journal device st_rdev */
+ __le32 jp_journal_size; /* size of the journal */
+ __le32 jp_journal_trans_max; /* max number of blocks in a transaction. */
+ __le32 jp_journal_magic; /* random value made on fs creation (this
+ * was sb_journal_block_count) */
+ __le32 jp_journal_max_batch; /* max number of blocks to batch into a
+ * trans */
+ __le32 jp_journal_max_commit_age; /* in seconds, how old can an async
+ * commit be */
+ __le32 jp_journal_max_trans_age; /* in seconds, how old can a transaction
+ * be */
+};
+
+/* this is the super from 3.5.X, where X >= 10 */
+struct reiserfs_super_block_v1 {
+ __le32 s_block_count; /* blocks count */
+ __le32 s_free_blocks; /* free blocks count */
+ __le32 s_root_block; /* root block number */
+ struct journal_params s_journal;
+ __le16 s_blocksize; /* block size */
+ __le16 s_oid_maxsize; /* max size of object id array, see
+ * get_objectid() commentary */
+ __le16 s_oid_cursize; /* current size of object id array */
+ __le16 s_umount_state; /* this is set to 1 when filesystem was
+ * umounted, to 2 - when not */
+ char s_magic[10]; /* reiserfs magic string indicates that
+ * file system is reiserfs:
+ * "ReIsErFs" or "ReIsEr2Fs" or "ReIsEr3Fs" */
+ __le16 s_fs_state; /* it is set to used by fsck to mark which
+ * phase of rebuilding is done */
+ __le32 s_hash_function_code; /* indicate, what hash function is being use
+ * to sort names in a directory*/
+ __le16 s_tree_height; /* height of disk tree */
+ __le16 s_bmap_nr; /* amount of bitmap blocks needed to address
+ * each block of file system */
+ __le16 s_version; /* this field is only reliable on filesystem
+ * with non-standard journal */
+ __le16 s_reserved_for_journal; /* size in blocks of journal area on main
+ * device, we need to keep after
+ * making fs with non-standard journal */
+} __attribute__ ((__packed__));
+
+#define SB_SIZE_V1 (sizeof(struct reiserfs_super_block_v1))
+
+/* this is the on disk super block */
+struct reiserfs_super_block {
+ struct reiserfs_super_block_v1 s_v1;
+ __le32 s_inode_generation;
+ __le32 s_flags; /* Right now used only by inode-attributes, if enabled */
+ unsigned char s_uuid[16]; /* filesystem unique identifier */
+ unsigned char s_label[16]; /* filesystem volume label */
+ __le16 s_mnt_count; /* Count of mounts since last fsck */
+ __le16 s_max_mnt_count; /* Maximum mounts before check */
+ __le32 s_lastcheck; /* Timestamp of last fsck */
+ __le32 s_check_interval; /* Interval between checks */
+ char s_unused[76]; /* zero filled by mkreiserfs and
+ * reiserfs_convert_objectid_map_v1()
+ * so any additions must be updated
+ * there as well. */
+} __attribute__ ((__packed__));
+
+#define SB_SIZE (sizeof(struct reiserfs_super_block))
+
+#define REISERFS_VERSION_1 0
+#define REISERFS_VERSION_2 2
+
+// on-disk super block fields converted to cpu form
+#define SB_DISK_SUPER_BLOCK(s) (REISERFS_SB(s)->s_rs)
+#define SB_V1_DISK_SUPER_BLOCK(s) (&(SB_DISK_SUPER_BLOCK(s)->s_v1))
+#define SB_BLOCKSIZE(s) \
+ le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_blocksize))
+#define SB_BLOCK_COUNT(s) \
+ le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_block_count))
+#define SB_FREE_BLOCKS(s) \
+ le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_free_blocks))
+#define SB_REISERFS_MAGIC(s) \
+ (SB_V1_DISK_SUPER_BLOCK(s)->s_magic)
+#define SB_ROOT_BLOCK(s) \
+ le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_root_block))
+#define SB_TREE_HEIGHT(s) \
+ le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_tree_height))
+#define SB_REISERFS_STATE(s) \
+ le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_umount_state))
+#define SB_VERSION(s) le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_version))
+#define SB_BMAP_NR(s) le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_bmap_nr))
+
+#define PUT_SB_BLOCK_COUNT(s, val) \
+ do { SB_V1_DISK_SUPER_BLOCK(s)->s_block_count = cpu_to_le32(val); } while (0)
+#define PUT_SB_FREE_BLOCKS(s, val) \
+ do { SB_V1_DISK_SUPER_BLOCK(s)->s_free_blocks = cpu_to_le32(val); } while (0)
+#define PUT_SB_ROOT_BLOCK(s, val) \
+ do { SB_V1_DISK_SUPER_BLOCK(s)->s_root_block = cpu_to_le32(val); } while (0)
+#define PUT_SB_TREE_HEIGHT(s, val) \
+ do { SB_V1_DISK_SUPER_BLOCK(s)->s_tree_height = cpu_to_le16(val); } while (0)
+#define PUT_SB_REISERFS_STATE(s, val) \
+ do { SB_V1_DISK_SUPER_BLOCK(s)->s_umount_state = cpu_to_le16(val); } while (0)
+#define PUT_SB_VERSION(s, val) \
+ do { SB_V1_DISK_SUPER_BLOCK(s)->s_version = cpu_to_le16(val); } while (0)
+#define PUT_SB_BMAP_NR(s, val) \
+ do { SB_V1_DISK_SUPER_BLOCK(s)->s_bmap_nr = cpu_to_le16 (val); } while (0)
+
+#define SB_ONDISK_JP(s) (&SB_V1_DISK_SUPER_BLOCK(s)->s_journal)
+#define SB_ONDISK_JOURNAL_SIZE(s) \
+ le32_to_cpu ((SB_ONDISK_JP(s)->jp_journal_size))
+#define SB_ONDISK_JOURNAL_1st_BLOCK(s) \
+ le32_to_cpu ((SB_ONDISK_JP(s)->jp_journal_1st_block))
+#define SB_ONDISK_JOURNAL_DEVICE(s) \
+ le32_to_cpu ((SB_ONDISK_JP(s)->jp_journal_dev))
+#define SB_ONDISK_RESERVED_FOR_JOURNAL(s) \
+ le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_reserved_for_journal))
+
+#define is_block_in_log_or_reserved_area(s, block) \
+ block >= SB_JOURNAL_1st_RESERVED_BLOCK(s) \
+ && block < SB_JOURNAL_1st_RESERVED_BLOCK(s) + \
+ ((!is_reiserfs_jr(SB_DISK_SUPER_BLOCK(s)) ? \
+ SB_ONDISK_JOURNAL_SIZE(s) + 1 : SB_ONDISK_RESERVED_FOR_JOURNAL(s)))
+
+int is_reiserfs_3_5(struct reiserfs_super_block *rs);
+int is_reiserfs_3_6(struct reiserfs_super_block *rs);
+int is_reiserfs_jr(struct reiserfs_super_block *rs);
+
+/* ReiserFS leaves the first 64k unused, so that partition labels have
+ enough space. If someone wants to write a fancy bootloader that
+ needs more than 64k, let us know, and this will be increased in size.
+ This number must be larger than than the largest block size on any
+ platform, or code will break. -Hans */
+#define REISERFS_DISK_OFFSET_IN_BYTES (64 * 1024)
+#define REISERFS_FIRST_BLOCK unused_define
+#define REISERFS_JOURNAL_OFFSET_IN_BYTES REISERFS_DISK_OFFSET_IN_BYTES
+
+/* the spot for the super in versions 3.5 - 3.5.10 (inclusive) */
+#define REISERFS_OLD_DISK_OFFSET_IN_BYTES (8 * 1024)
+
+/* reiserfs internal error code (used by search_by_key and fix_nodes)) */
+#define CARRY_ON 0
+#define REPEAT_SEARCH -1
+#define IO_ERROR -2
+#define NO_DISK_SPACE -3
+#define NO_BALANCING_NEEDED (-4)
+#define NO_MORE_UNUSED_CONTIGUOUS_BLOCKS (-5)
+#define QUOTA_EXCEEDED -6
+
+typedef __u32 b_blocknr_t;
+typedef __le32 unp_t;
+
+struct unfm_nodeinfo {
+ unp_t unfm_nodenum;
+ unsigned short unfm_freespace;
+};
+
+/* there are two formats of keys: 3.5 and 3.6
+ */
+#define KEY_FORMAT_3_5 0
+#define KEY_FORMAT_3_6 1
+
+/* there are two stat datas */
+#define STAT_DATA_V1 0
+#define STAT_DATA_V2 1
+
+static inline struct reiserfs_inode_info *REISERFS_I(const struct inode *inode)
+{
+ return container_of(inode, struct reiserfs_inode_info, vfs_inode);
+}
+
+static inline struct reiserfs_sb_info *REISERFS_SB(const struct super_block *sb)
+{
+ return sb->s_fs_info;
+}
+
+/* Don't trust REISERFS_SB(sb)->s_bmap_nr, it's a u16
+ * which overflows on large file systems. */
+static inline __u32 reiserfs_bmap_count(struct super_block *sb)
+{
+ return (SB_BLOCK_COUNT(sb) - 1) / (sb->s_blocksize * 8) + 1;
+}
+
+static inline int bmap_would_wrap(unsigned bmap_nr)
+{
+ return bmap_nr > ((1LL << 16) - 1);
+}
+
+/** this says about version of key of all items (but stat data) the
+ object consists of */
+#define get_inode_item_key_version( inode ) \
+ ((REISERFS_I(inode)->i_flags & i_item_key_version_mask) ? KEY_FORMAT_3_6 : KEY_FORMAT_3_5)
+
+#define set_inode_item_key_version( inode, version ) \
+ ({ if((version)==KEY_FORMAT_3_6) \
+ REISERFS_I(inode)->i_flags |= i_item_key_version_mask; \
+ else \
+ REISERFS_I(inode)->i_flags &= ~i_item_key_version_mask; })
+
+#define get_inode_sd_version(inode) \
+ ((REISERFS_I(inode)->i_flags & i_stat_data_version_mask) ? STAT_DATA_V2 : STAT_DATA_V1)
+
+#define set_inode_sd_version(inode, version) \
+ ({ if((version)==STAT_DATA_V2) \
+ REISERFS_I(inode)->i_flags |= i_stat_data_version_mask; \
+ else \
+ REISERFS_I(inode)->i_flags &= ~i_stat_data_version_mask; })
+
+/* This is an aggressive tail suppression policy, I am hoping it
+ improves our benchmarks. The principle behind it is that percentage
+ space saving is what matters, not absolute space saving. This is
+ non-intuitive, but it helps to understand it if you consider that the
+ cost to access 4 blocks is not much more than the cost to access 1
+ block, if you have to do a seek and rotate. A tail risks a
+ non-linear disk access that is significant as a percentage of total
+ time cost for a 4 block file and saves an amount of space that is
+ less significant as a percentage of space, or so goes the hypothesis.
+ -Hans */
+#define STORE_TAIL_IN_UNFM_S1(n_file_size,n_tail_size,n_block_size) \
+(\
+ (!(n_tail_size)) || \
+ (((n_tail_size) > MAX_DIRECT_ITEM_LEN(n_block_size)) || \
+ ( (n_file_size) >= (n_block_size) * 4 ) || \
+ ( ( (n_file_size) >= (n_block_size) * 3 ) && \
+ ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size))/4) ) || \
+ ( ( (n_file_size) >= (n_block_size) * 2 ) && \
+ ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size))/2) ) || \
+ ( ( (n_file_size) >= (n_block_size) ) && \
+ ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size) * 3)/4) ) ) \
+)
+
+/* Another strategy for tails, this one means only create a tail if all the
+ file would fit into one DIRECT item.
+ Primary intention for this one is to increase performance by decreasing
+ seeking.
+*/
+#define STORE_TAIL_IN_UNFM_S2(n_file_size,n_tail_size,n_block_size) \
+(\
+ (!(n_tail_size)) || \
+ (((n_file_size) > MAX_DIRECT_ITEM_LEN(n_block_size)) ) \
+)
+
+/*
+ * values for s_umount_state field
+ */
+#define REISERFS_VALID_FS 1
+#define REISERFS_ERROR_FS 2
+
+//
+// there are 5 item types currently
+//
+#define TYPE_STAT_DATA 0
+#define TYPE_INDIRECT 1
+#define TYPE_DIRECT 2
+#define TYPE_DIRENTRY 3
+#define TYPE_MAXTYPE 3
+#define TYPE_ANY 15 // FIXME: comment is required
+
+/***************************************************************************/
+/* KEY & ITEM HEAD */
+/***************************************************************************/
+
+//
+// directories use this key as well as old files
+//
+struct offset_v1 {
+ __le32 k_offset;
+ __le32 k_uniqueness;
+} __attribute__ ((__packed__));
+
+struct offset_v2 {
+ __le64 v;
+} __attribute__ ((__packed__));
+
+static inline __u16 offset_v2_k_type(const struct offset_v2 *v2)
+{
+ __u8 type = le64_to_cpu(v2->v) >> 60;
+ return (type <= TYPE_MAXTYPE) ? type : TYPE_ANY;
+}
+
+static inline void set_offset_v2_k_type(struct offset_v2 *v2, int type)
+{
+ v2->v =
+ (v2->v & cpu_to_le64(~0ULL >> 4)) | cpu_to_le64((__u64) type << 60);
+}
+
+static inline loff_t offset_v2_k_offset(const struct offset_v2 *v2)
+{
+ return le64_to_cpu(v2->v) & (~0ULL >> 4);
+}
+
+static inline void set_offset_v2_k_offset(struct offset_v2 *v2, loff_t offset)
+{
+ offset &= (~0ULL >> 4);
+ v2->v = (v2->v & cpu_to_le64(15ULL << 60)) | cpu_to_le64(offset);
+}
+
+/* Key of an item determines its location in the S+tree, and
+ is composed of 4 components */
+struct reiserfs_key {
+ __le32 k_dir_id; /* packing locality: by default parent
+ directory object id */
+ __le32 k_objectid; /* object identifier */
+ union {
+ struct offset_v1 k_offset_v1;
+ struct offset_v2 k_offset_v2;
+ } __attribute__ ((__packed__)) u;
+} __attribute__ ((__packed__));
+
+struct in_core_key {
+ __u32 k_dir_id; /* packing locality: by default parent
+ directory object id */
+ __u32 k_objectid; /* object identifier */
+ __u64 k_offset;
+ __u8 k_type;
+};
+
+struct cpu_key {
+ struct in_core_key on_disk_key;
+ int version;
+ int key_length; /* 3 in all cases but direct2indirect and
+ indirect2direct conversion */
+};
+
+/* Our function for comparing keys can compare keys of different
+ lengths. It takes as a parameter the length of the keys it is to
+ compare. These defines are used in determining what is to be passed
+ to it as that parameter. */
+#define REISERFS_FULL_KEY_LEN 4
+#define REISERFS_SHORT_KEY_LEN 2
+
+/* The result of the key compare */
+#define FIRST_GREATER 1
+#define SECOND_GREATER -1
+#define KEYS_IDENTICAL 0
+#define KEY_FOUND 1
+#define KEY_NOT_FOUND 0
+
+#define KEY_SIZE (sizeof(struct reiserfs_key))
+#define SHORT_KEY_SIZE (sizeof (__u32) + sizeof (__u32))
+
+/* return values for search_by_key and clones */
+#define ITEM_FOUND 1
+#define ITEM_NOT_FOUND 0
+#define ENTRY_FOUND 1
+#define ENTRY_NOT_FOUND 0
+#define DIRECTORY_NOT_FOUND -1
+#define REGULAR_FILE_FOUND -2
+#define DIRECTORY_FOUND -3
+#define BYTE_FOUND 1
+#define BYTE_NOT_FOUND 0
+#define FILE_NOT_FOUND -1
+
+#define POSITION_FOUND 1
+#define POSITION_NOT_FOUND 0
+
+// return values for reiserfs_find_entry and search_by_entry_key
+#define NAME_FOUND 1
+#define NAME_NOT_FOUND 0
+#define GOTO_PREVIOUS_ITEM 2
+#define NAME_FOUND_INVISIBLE 3
+
+/* Everything in the filesystem is stored as a set of items. The
+ item head contains the key of the item, its free space (for
+ indirect items) and specifies the location of the item itself
+ within the block. */
+
+struct item_head {
+ /* Everything in the tree is found by searching for it based on
+ * its key.*/
+ struct reiserfs_key ih_key;
+ union {
+ /* The free space in the last unformatted node of an
+ indirect item if this is an indirect item. This
+ equals 0xFFFF iff this is a direct item or stat data
+ item. Note that the key, not this field, is used to
+ determine the item type, and thus which field this
+ union contains. */
+ __le16 ih_free_space_reserved;
+ /* Iff this is a directory item, this field equals the
+ number of directory entries in the directory item. */
+ __le16 ih_entry_count;
+ } __attribute__ ((__packed__)) u;
+ __le16 ih_item_len; /* total size of the item body */
+ __le16 ih_item_location; /* an offset to the item body
+ * within the block */
+ __le16 ih_version; /* 0 for all old items, 2 for new
+ ones. Highest bit is set by fsck
+ temporary, cleaned after all
+ done */
+} __attribute__ ((__packed__));
+/* size of item header */
+#define IH_SIZE (sizeof(struct item_head))
+
+#define ih_free_space(ih) le16_to_cpu((ih)->u.ih_free_space_reserved)
+#define ih_version(ih) le16_to_cpu((ih)->ih_version)
+#define ih_entry_count(ih) le16_to_cpu((ih)->u.ih_entry_count)
+#define ih_location(ih) le16_to_cpu((ih)->ih_item_location)
+#define ih_item_len(ih) le16_to_cpu((ih)->ih_item_len)
+
+#define put_ih_free_space(ih, val) do { (ih)->u.ih_free_space_reserved = cpu_to_le16(val); } while(0)
+#define put_ih_version(ih, val) do { (ih)->ih_version = cpu_to_le16(val); } while (0)
+#define put_ih_entry_count(ih, val) do { (ih)->u.ih_entry_count = cpu_to_le16(val); } while (0)
+#define put_ih_location(ih, val) do { (ih)->ih_item_location = cpu_to_le16(val); } while (0)
+#define put_ih_item_len(ih, val) do { (ih)->ih_item_len = cpu_to_le16(val); } while (0)
+
+#define unreachable_item(ih) (ih_version(ih) & (1 << 15))
+
+#define get_ih_free_space(ih) (ih_version (ih) == KEY_FORMAT_3_6 ? 0 : ih_free_space (ih))
+#define set_ih_free_space(ih,val) put_ih_free_space((ih), ((ih_version(ih) == KEY_FORMAT_3_6) ? 0 : (val)))
+
+/* these operate on indirect items, where you've got an array of ints
+** at a possibly unaligned location. These are a noop on ia32
+**
+** p is the array of __u32, i is the index into the array, v is the value
+** to store there.
+*/
+#define get_block_num(p, i) get_unaligned_le32((p) + (i))
+#define put_block_num(p, i, v) put_unaligned_le32((v), (p) + (i))
+
+//
+// in old version uniqueness field shows key type
+//
+#define V1_SD_UNIQUENESS 0
+#define V1_INDIRECT_UNIQUENESS 0xfffffffe
+#define V1_DIRECT_UNIQUENESS 0xffffffff
+#define V1_DIRENTRY_UNIQUENESS 500
+#define V1_ANY_UNIQUENESS 555 // FIXME: comment is required
+
+//
+// here are conversion routines
+//
+static inline int uniqueness2type(__u32 uniqueness) CONSTF;
+static inline int uniqueness2type(__u32 uniqueness)
+{
+ switch ((int)uniqueness) {
+ case V1_SD_UNIQUENESS:
+ return TYPE_STAT_DATA;
+ case V1_INDIRECT_UNIQUENESS:
+ return TYPE_INDIRECT;
+ case V1_DIRECT_UNIQUENESS:
+ return TYPE_DIRECT;
+ case V1_DIRENTRY_UNIQUENESS:
+ return TYPE_DIRENTRY;
+ case V1_ANY_UNIQUENESS:
+ default:
+ return TYPE_ANY;
+ }
+}
+
+static inline __u32 type2uniqueness(int type) CONSTF;
+static inline __u32 type2uniqueness(int type)
+{
+ switch (type) {
+ case TYPE_STAT_DATA:
+ return V1_SD_UNIQUENESS;
+ case TYPE_INDIRECT:
+ return V1_INDIRECT_UNIQUENESS;
+ case TYPE_DIRECT:
+ return V1_DIRECT_UNIQUENESS;
+ case TYPE_DIRENTRY:
+ return V1_DIRENTRY_UNIQUENESS;
+ case TYPE_ANY:
+ default:
+ return V1_ANY_UNIQUENESS;
+ }
+}
+
+//
+// key is pointer to on disk key which is stored in le, result is cpu,
+// there is no way to get version of object from key, so, provide
+// version to these defines
+//
+static inline loff_t le_key_k_offset(int version,
+ const struct reiserfs_key *key)
+{
+ return (version == KEY_FORMAT_3_5) ?
+ le32_to_cpu(key->u.k_offset_v1.k_offset) :
+ offset_v2_k_offset(&(key->u.k_offset_v2));
+}
+
+static inline loff_t le_ih_k_offset(const struct item_head *ih)
+{
+ return le_key_k_offset(ih_version(ih), &(ih->ih_key));
+}
+
+static inline loff_t le_key_k_type(int version, const struct reiserfs_key *key)
+{
+ return (version == KEY_FORMAT_3_5) ?
+ uniqueness2type(le32_to_cpu(key->u.k_offset_v1.k_uniqueness)) :
+ offset_v2_k_type(&(key->u.k_offset_v2));
+}
+
+static inline loff_t le_ih_k_type(const struct item_head *ih)
+{
+ return le_key_k_type(ih_version(ih), &(ih->ih_key));
+}
+
+static inline void set_le_key_k_offset(int version, struct reiserfs_key *key,
+ loff_t offset)
+{
+ (version == KEY_FORMAT_3_5) ? (void)(key->u.k_offset_v1.k_offset = cpu_to_le32(offset)) : /* jdm check */
+ (void)(set_offset_v2_k_offset(&(key->u.k_offset_v2), offset));
+}
+
+static inline void set_le_ih_k_offset(struct item_head *ih, loff_t offset)
+{
+ set_le_key_k_offset(ih_version(ih), &(ih->ih_key), offset);
+}
+
+static inline void set_le_key_k_type(int version, struct reiserfs_key *key,
+ int type)
+{
+ (version == KEY_FORMAT_3_5) ?
+ (void)(key->u.k_offset_v1.k_uniqueness =
+ cpu_to_le32(type2uniqueness(type)))
+ : (void)(set_offset_v2_k_type(&(key->u.k_offset_v2), type));
+}
+
+static inline void set_le_ih_k_type(struct item_head *ih, int type)
+{
+ set_le_key_k_type(ih_version(ih), &(ih->ih_key), type);
+}
+
+static inline int is_direntry_le_key(int version, struct reiserfs_key *key)
+{
+ return le_key_k_type(version, key) == TYPE_DIRENTRY;
+}
+
+static inline int is_direct_le_key(int version, struct reiserfs_key *key)
+{
+ return le_key_k_type(version, key) == TYPE_DIRECT;
+}
+
+static inline int is_indirect_le_key(int version, struct reiserfs_key *key)
+{
+ return le_key_k_type(version, key) == TYPE_INDIRECT;
+}
+
+static inline int is_statdata_le_key(int version, struct reiserfs_key *key)
+{
+ return le_key_k_type(version, key) == TYPE_STAT_DATA;
+}
+
+//
+// item header has version.
+//
+static inline int is_direntry_le_ih(struct item_head *ih)
+{
+ return is_direntry_le_key(ih_version(ih), &ih->ih_key);
+}
+
+static inline int is_direct_le_ih(struct item_head *ih)
+{
+ return is_direct_le_key(ih_version(ih), &ih->ih_key);
+}
+
+static inline int is_indirect_le_ih(struct item_head *ih)
+{
+ return is_indirect_le_key(ih_version(ih), &ih->ih_key);
+}
+
+static inline int is_statdata_le_ih(struct item_head *ih)
+{
+ return is_statdata_le_key(ih_version(ih), &ih->ih_key);
+}
+
+//
+// key is pointer to cpu key, result is cpu
+//
+static inline loff_t cpu_key_k_offset(const struct cpu_key *key)
+{
+ return key->on_disk_key.k_offset;
+}
+
+static inline loff_t cpu_key_k_type(const struct cpu_key *key)
+{
+ return key->on_disk_key.k_type;
+}
+
+static inline void set_cpu_key_k_offset(struct cpu_key *key, loff_t offset)
+{
+ key->on_disk_key.k_offset = offset;
+}
+
+static inline void set_cpu_key_k_type(struct cpu_key *key, int type)
+{
+ key->on_disk_key.k_type = type;
+}
+
+static inline void cpu_key_k_offset_dec(struct cpu_key *key)
+{
+ key->on_disk_key.k_offset--;
+}
+
+#define is_direntry_cpu_key(key) (cpu_key_k_type (key) == TYPE_DIRENTRY)
+#define is_direct_cpu_key(key) (cpu_key_k_type (key) == TYPE_DIRECT)
+#define is_indirect_cpu_key(key) (cpu_key_k_type (key) == TYPE_INDIRECT)
+#define is_statdata_cpu_key(key) (cpu_key_k_type (key) == TYPE_STAT_DATA)
+
+/* are these used ? */
+#define is_direntry_cpu_ih(ih) (is_direntry_cpu_key (&((ih)->ih_key)))
+#define is_direct_cpu_ih(ih) (is_direct_cpu_key (&((ih)->ih_key)))
+#define is_indirect_cpu_ih(ih) (is_indirect_cpu_key (&((ih)->ih_key)))
+#define is_statdata_cpu_ih(ih) (is_statdata_cpu_key (&((ih)->ih_key)))
+
+#define I_K_KEY_IN_ITEM(ih, key, n_blocksize) \
+ (!COMP_SHORT_KEYS(ih, key) && \
+ I_OFF_BYTE_IN_ITEM(ih, k_offset(key), n_blocksize))
+
+/* maximal length of item */
+#define MAX_ITEM_LEN(block_size) (block_size - BLKH_SIZE - IH_SIZE)
+#define MIN_ITEM_LEN 1
+
+/* object identifier for root dir */
+#define REISERFS_ROOT_OBJECTID 2
+#define REISERFS_ROOT_PARENT_OBJECTID 1
+
+extern struct reiserfs_key root_key;
+
+/*
+ * Picture represents a leaf of the S+tree
+ * ______________________________________________________
+ * | | Array of | | |
+ * |Block | Object-Item | F r e e | Objects- |
+ * | head | Headers | S p a c e | Items |
+ * |______|_______________|___________________|___________|
+ */
+
+/* Header of a disk block. More precisely, header of a formatted leaf
+ or internal node, and not the header of an unformatted node. */
+struct block_head {
+ __le16 blk_level; /* Level of a block in the tree. */
+ __le16 blk_nr_item; /* Number of keys/items in a block. */
+ __le16 blk_free_space; /* Block free space in bytes. */
+ __le16 blk_reserved;
+ /* dump this in v4/planA */
+ struct reiserfs_key blk_right_delim_key; /* kept only for compatibility */
+};
+
+#define BLKH_SIZE (sizeof(struct block_head))
+#define blkh_level(p_blkh) (le16_to_cpu((p_blkh)->blk_level))
+#define blkh_nr_item(p_blkh) (le16_to_cpu((p_blkh)->blk_nr_item))
+#define blkh_free_space(p_blkh) (le16_to_cpu((p_blkh)->blk_free_space))
+#define blkh_reserved(p_blkh) (le16_to_cpu((p_blkh)->blk_reserved))
+#define set_blkh_level(p_blkh,val) ((p_blkh)->blk_level = cpu_to_le16(val))
+#define set_blkh_nr_item(p_blkh,val) ((p_blkh)->blk_nr_item = cpu_to_le16(val))
+#define set_blkh_free_space(p_blkh,val) ((p_blkh)->blk_free_space = cpu_to_le16(val))
+#define set_blkh_reserved(p_blkh,val) ((p_blkh)->blk_reserved = cpu_to_le16(val))
+#define blkh_right_delim_key(p_blkh) ((p_blkh)->blk_right_delim_key)
+#define set_blkh_right_delim_key(p_blkh,val) ((p_blkh)->blk_right_delim_key = val)
+
+/*
+ * values for blk_level field of the struct block_head
+ */
+
+#define FREE_LEVEL 0 /* when node gets removed from the tree its
+ blk_level is set to FREE_LEVEL. It is then
+ used to see whether the node is still in the
+ tree */
+
+#define DISK_LEAF_NODE_LEVEL 1 /* Leaf node level. */
+
+/* Given the buffer head of a formatted node, resolve to the block head of that node. */
+#define B_BLK_HEAD(bh) ((struct block_head *)((bh)->b_data))
+/* Number of items that are in buffer. */
+#define B_NR_ITEMS(bh) (blkh_nr_item(B_BLK_HEAD(bh)))
+#define B_LEVEL(bh) (blkh_level(B_BLK_HEAD(bh)))
+#define B_FREE_SPACE(bh) (blkh_free_space(B_BLK_HEAD(bh)))
+
+#define PUT_B_NR_ITEMS(bh, val) do { set_blkh_nr_item(B_BLK_HEAD(bh), val); } while (0)
+#define PUT_B_LEVEL(bh, val) do { set_blkh_level(B_BLK_HEAD(bh), val); } while (0)
+#define PUT_B_FREE_SPACE(bh, val) do { set_blkh_free_space(B_BLK_HEAD(bh), val); } while (0)
+
+/* Get right delimiting key. -- little endian */
+#define B_PRIGHT_DELIM_KEY(bh) (&(blk_right_delim_key(B_BLK_HEAD(bh))))
+
+/* Does the buffer contain a disk leaf. */
+#define B_IS_ITEMS_LEVEL(bh) (B_LEVEL(bh) == DISK_LEAF_NODE_LEVEL)
+
+/* Does the buffer contain a disk internal node */
+#define B_IS_KEYS_LEVEL(bh) (B_LEVEL(bh) > DISK_LEAF_NODE_LEVEL \
+ && B_LEVEL(bh) <= MAX_HEIGHT)
+
+/***************************************************************************/
+/* STAT DATA */
+/***************************************************************************/
+
+//
+// old stat data is 32 bytes long. We are going to distinguish new one by
+// different size
+//
+struct stat_data_v1 {
+ __le16 sd_mode; /* file type, permissions */
+ __le16 sd_nlink; /* number of hard links */
+ __le16 sd_uid; /* owner */
+ __le16 sd_gid; /* group */
+ __le32 sd_size; /* file size */
+ __le32 sd_atime; /* time of last access */
+ __le32 sd_mtime; /* time file was last modified */
+ __le32 sd_ctime; /* time inode (stat data) was last changed (except changes to sd_atime and sd_mtime) */
+ union {
+ __le32 sd_rdev;
+ __le32 sd_blocks; /* number of blocks file uses */
+ } __attribute__ ((__packed__)) u;
+ __le32 sd_first_direct_byte; /* first byte of file which is stored
+ in a direct item: except that if it
+ equals 1 it is a symlink and if it
+ equals ~(__u32)0 there is no
+ direct item. The existence of this
+ field really grates on me. Let's
+ replace it with a macro based on
+ sd_size and our tail suppression
+ policy. Someday. -Hans */
+} __attribute__ ((__packed__));
+
+#define SD_V1_SIZE (sizeof(struct stat_data_v1))
+#define stat_data_v1(ih) (ih_version (ih) == KEY_FORMAT_3_5)
+#define sd_v1_mode(sdp) (le16_to_cpu((sdp)->sd_mode))
+#define set_sd_v1_mode(sdp,v) ((sdp)->sd_mode = cpu_to_le16(v))
+#define sd_v1_nlink(sdp) (le16_to_cpu((sdp)->sd_nlink))
+#define set_sd_v1_nlink(sdp,v) ((sdp)->sd_nlink = cpu_to_le16(v))
+#define sd_v1_uid(sdp) (le16_to_cpu((sdp)->sd_uid))
+#define set_sd_v1_uid(sdp,v) ((sdp)->sd_uid = cpu_to_le16(v))
+#define sd_v1_gid(sdp) (le16_to_cpu((sdp)->sd_gid))
+#define set_sd_v1_gid(sdp,v) ((sdp)->sd_gid = cpu_to_le16(v))
+#define sd_v1_size(sdp) (le32_to_cpu((sdp)->sd_size))
+#define set_sd_v1_size(sdp,v) ((sdp)->sd_size = cpu_to_le32(v))
+#define sd_v1_atime(sdp) (le32_to_cpu((sdp)->sd_atime))
+#define set_sd_v1_atime(sdp,v) ((sdp)->sd_atime = cpu_to_le32(v))
+#define sd_v1_mtime(sdp) (le32_to_cpu((sdp)->sd_mtime))
+#define set_sd_v1_mtime(sdp,v) ((sdp)->sd_mtime = cpu_to_le32(v))
+#define sd_v1_ctime(sdp) (le32_to_cpu((sdp)->sd_ctime))
+#define set_sd_v1_ctime(sdp,v) ((sdp)->sd_ctime = cpu_to_le32(v))
+#define sd_v1_rdev(sdp) (le32_to_cpu((sdp)->u.sd_rdev))
+#define set_sd_v1_rdev(sdp,v) ((sdp)->u.sd_rdev = cpu_to_le32(v))
+#define sd_v1_blocks(sdp) (le32_to_cpu((sdp)->u.sd_blocks))
+#define set_sd_v1_blocks(sdp,v) ((sdp)->u.sd_blocks = cpu_to_le32(v))
+#define sd_v1_first_direct_byte(sdp) \
+ (le32_to_cpu((sdp)->sd_first_direct_byte))
+#define set_sd_v1_first_direct_byte(sdp,v) \
+ ((sdp)->sd_first_direct_byte = cpu_to_le32(v))
+
+/* inode flags stored in sd_attrs (nee sd_reserved) */
+
+/* we want common flags to have the same values as in ext2,
+ so chattr(1) will work without problems */
+#define REISERFS_IMMUTABLE_FL FS_IMMUTABLE_FL
+#define REISERFS_APPEND_FL FS_APPEND_FL
+#define REISERFS_SYNC_FL FS_SYNC_FL
+#define REISERFS_NOATIME_FL FS_NOATIME_FL
+#define REISERFS_NODUMP_FL FS_NODUMP_FL
+#define REISERFS_SECRM_FL FS_SECRM_FL
+#define REISERFS_UNRM_FL FS_UNRM_FL
+#define REISERFS_COMPR_FL FS_COMPR_FL
+#define REISERFS_NOTAIL_FL FS_NOTAIL_FL
+
+/* persistent flags that file inherits from the parent directory */
+#define REISERFS_INHERIT_MASK ( REISERFS_IMMUTABLE_FL | \
+ REISERFS_SYNC_FL | \
+ REISERFS_NOATIME_FL | \
+ REISERFS_NODUMP_FL | \
+ REISERFS_SECRM_FL | \
+ REISERFS_COMPR_FL | \
+ REISERFS_NOTAIL_FL )
+
+/* Stat Data on disk (reiserfs version of UFS disk inode minus the
+ address blocks) */
+struct stat_data {
+ __le16 sd_mode; /* file type, permissions */
+ __le16 sd_attrs; /* persistent inode flags */
+ __le32 sd_nlink; /* number of hard links */
+ __le64 sd_size; /* file size */
+ __le32 sd_uid; /* owner */
+ __le32 sd_gid; /* group */
+ __le32 sd_atime; /* time of last access */
+ __le32 sd_mtime; /* time file was last modified */
+ __le32 sd_ctime; /* time inode (stat data) was last changed (except changes to sd_atime and sd_mtime) */
+ __le32 sd_blocks;
+ union {
+ __le32 sd_rdev;
+ __le32 sd_generation;
+ //__le32 sd_first_direct_byte;
+ /* first byte of file which is stored in a
+ direct item: except that if it equals 1
+ it is a symlink and if it equals
+ ~(__u32)0 there is no direct item. The
+ existence of this field really grates
+ on me. Let's replace it with a macro
+ based on sd_size and our tail
+ suppression policy? */
+ } __attribute__ ((__packed__)) u;
+} __attribute__ ((__packed__));
+//
+// this is 44 bytes long
+//
+#define SD_SIZE (sizeof(struct stat_data))
+#define SD_V2_SIZE SD_SIZE
+#define stat_data_v2(ih) (ih_version (ih) == KEY_FORMAT_3_6)
+#define sd_v2_mode(sdp) (le16_to_cpu((sdp)->sd_mode))
+#define set_sd_v2_mode(sdp,v) ((sdp)->sd_mode = cpu_to_le16(v))
+/* sd_reserved */
+/* set_sd_reserved */
+#define sd_v2_nlink(sdp) (le32_to_cpu((sdp)->sd_nlink))
+#define set_sd_v2_nlink(sdp,v) ((sdp)->sd_nlink = cpu_to_le32(v))
+#define sd_v2_size(sdp) (le64_to_cpu((sdp)->sd_size))
+#define set_sd_v2_size(sdp,v) ((sdp)->sd_size = cpu_to_le64(v))
+#define sd_v2_uid(sdp) (le32_to_cpu((sdp)->sd_uid))
+#define set_sd_v2_uid(sdp,v) ((sdp)->sd_uid = cpu_to_le32(v))
+#define sd_v2_gid(sdp) (le32_to_cpu((