diff options
Diffstat (limited to 'fs/reiserfs/reiserfs.h')
| -rw-r--r-- | fs/reiserfs/reiserfs.h | 2923 |
1 files changed, 2923 insertions, 0 deletions
diff --git a/fs/reiserfs/reiserfs.h b/fs/reiserfs/reiserfs.h new file mode 100644 index 000000000000..a59d27126338 --- /dev/null +++ b/fs/reiserfs/reiserfs.h @@ -0,0 +1,2923 @@ +/* + * 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/bug.h> +#include <linux/workqueue.h> +#include <asm/unaligned.h> +#include <linux/bitops.h> +#include <linux/proc_fs.h> +#include <linux/buffer_head.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 + +struct reiserfs_journal_list; + +/** bitmasks for i_flags field in reiserfs-specific part of inode */ +typedef enum { + /** this says what format of key do all items (but stat data) of + an object have. If this is set, that format is 3.6 otherwise + - 3.5 */ + i_item_key_version_mask = 0x0001, + /** If this is unset, object has 3.5 stat data, otherwise, it has + 3.6 stat data with 64bit size, 32bit nlink etc. */ + i_stat_data_version_mask = 0x0002, + /** file might need tail packing on close */ + i_pack_on_close_mask = 0x0004, + /** don't pack tail of file */ + i_nopack_mask = 0x0008, + /** If those is set, "safe link" was created for this file during + truncate or unlink. Safe link is used to avoid leakage of disk + space on crash with some files open, but unlinked. */ + i_link_saved_unlink_mask = 0x0010, + i_link_saved_truncate_mask = 0x0020, + i_has_xattr_dir = 0x0040, + i_data_log = 0x0080, +} reiserfs_inode_flags; + +struct reiserfs_inode_info { + __u32 i_key[4]; /* key is still 4 32 bit integers */ + /** transient inode flags that are never stored on disk. Bitmasks + for this field are defined above. */ + __u32 i_flags; + + __u32 i_first_direct_byte; // offset of first byte stored in direct item. + + /* copy of persistent inode flags read from sd_attrs. */ + __u32 i_attrs; + + int i_prealloc_block; /* first unused block of a sequence of unused blocks */ + int i_prealloc_count; /* length of that sequence */ + struct list_head i_prealloc_list; /* per-transaction list of inodes which + * have preallocated blocks */ + + unsigned new_packing_locality:1; /* new_packig_locality is created; new blocks + * for the contents of this directory should be + * displaced */ + + /* we use these for fsync or O_SYNC to decide which transaction + ** needs to be committed in order for this inode to be properly + ** flushed */ + unsigned int i_trans_id; + struct reiserfs_journal_list *i_jl; + atomic_t openers; + struct mutex tailpack; +#ifdef CONFIG_REISERFS_FS_XATTR + struct rw_semaphore i_xattr_sem; +#endif + struct inode vfs_inode; +}; + +typedef enum { + reiserfs_attrs_cleared = 0x00000001, +} reiserfs_super_block_flags; + +/* struct reiserfs_super_block accessors/mutators + * since this is a disk structure, it will always be in + * little endian format. */ +#define sb_block_count(sbp) (le32_to_cpu((sbp)->s_v1.s_block_count)) +#define set_sb_block_count(sbp,v) ((sbp)->s_v1.s_block_count = cpu_to_le32(v)) +#define sb_free_blocks(sbp) (le32_to_cpu((sbp)->s_v1.s_free_blocks)) +#define set_sb_free_blocks(sbp,v) ((sbp)->s_v1.s_free_blocks = cpu_to_le32(v)) +#define sb_root_block(sbp) (le32_to_cpu((sbp)->s_v1.s_root_block)) +#define set_sb_root_block(sbp,v) ((sbp)->s_v1.s_root_block = cpu_to_le32(v)) + +#define sb_jp_journal_1st_block(sbp) \ + (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_1st_block)) +#define set_sb_jp_journal_1st_block(sbp,v) \ + ((sbp)->s_v1.s_journal.jp_journal_1st_block = cpu_to_le32(v)) +#define sb_jp_journal_dev(sbp) \ + (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_dev)) +#define set_sb_jp_journal_dev(sbp,v) \ + ((sbp)->s_v1.s_journal.jp_journal_dev = cpu_to_le32(v)) +#define sb_jp_journal_size(sbp) \ + (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_size)) +#define set_sb_jp_journal_size(sbp,v) \ + ((sbp)->s_v1.s_journal.jp_journal_size = cpu_to_le32(v)) +#define sb_jp_journal_trans_max(sbp) \ + (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_trans_max)) +#define set_sb_jp_journal_trans_max(sbp,v) \ + ((sbp)->s_v1.s_journal.jp_journal_trans_max = cpu_to_le32(v)) +#define sb_jp_journal_magic(sbp) \ + (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_magic)) +#define set_sb_jp_journal_magic(sbp,v) \ + ((sbp)->s_v1.s_journal.jp_journal_magic = cpu_to_le32(v)) +#define sb_jp_journal_max_batch(sbp) \ + (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_max_batch)) +#define set_sb_jp_journal_max_batch(sbp,v) \ + ((sbp)->s_v1.s_journal.jp_journal_max_batch = cpu_to_le32(v)) +#define sb_jp_jourmal_max_commit_age(sbp) \ + (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_max_commit_age)) +#define set_sb_jp_journal_max_commit_age(sbp,v) \ + ((sbp)->s_v1.s_journal.jp_journal_max_commit_age = cpu_to_le32(v)) + +#define sb_blocksize(sbp) (le16_to_cpu((sbp)->s_v1.s_blocksize)) +#define set_sb_blocksize(sbp,v) ((sbp)->s_v1.s_blocksize = cpu_to_le16(v)) +#define sb_oid_maxsize(sbp) (le16_to_cpu((sbp)->s_v1.s_oid_maxsize)) +#define set_sb_oid_maxsize(sbp,v) ((sbp)->s_v1.s_oid_maxsize = cpu_to_le16(v)) +#define sb_oid_cursize(sbp) (le16_to_cpu((sbp)->s_v1.s_oid_cursize)) +#define set_sb_oid_cursize(sbp,v) ((sbp)->s_v1.s_oid_cursize = cpu_to_le16(v)) +#define sb_umount_state(sbp) (le16_to_cpu((sbp)->s_v1.s_umount_state)) +#define set_sb_umount_state(sbp,v) ((sbp)->s_v1.s_umount_state = cpu_to_le16(v)) +#define sb_fs_state(sbp) (le16_to_cpu((sbp)->s_v1.s_fs_state)) +#define set_sb_fs_state(sbp,v) ((sbp)->s_v1.s_fs_state = cpu_to_le16(v)) +#define sb_hash_function_code(sbp) \ + (le32_to_cpu((sbp)->s_v1.s_hash_function_code)) +#define set_sb_hash_function_code(sbp,v) \ + ((sbp)->s_v1.s_hash_function_code = cpu_to_le32(v)) +#define sb_tree_height(sbp) (le16_to_cpu((sbp)->s_v1.s_tree_height)) +#define set_sb_tree_height(sbp,v) ((sbp)->s_v1.s_tree_height = cpu_to_le16(v)) +#define sb_bmap_nr(sbp) (le16_to_cpu((sbp)->s_v1.s_bmap_nr)) +#define set_sb_bmap_nr(sbp,v) ((sbp)->s_v1.s_bmap_nr = cpu_to_le16(v)) +#define sb_version(sbp) (le16_to_cpu((sbp)->s_v1.s_version)) +#define set_sb_version(sbp,v) ((sbp)->s_v1.s_version = cpu_to_le16(v)) + +#define sb_mnt_count(sbp) (le16_to_cpu((sbp)->s_mnt_count)) +#define set_sb_mnt_count(sbp, v) ((sbp)->s_mnt_count = cpu_to_le16(v)) + +#define sb_reserved_for_journal(sbp) \ + (le16_to_cpu((sbp)->s_v1.s_reserved_for_journal)) +#define set_sb_reserved_for_journal(sbp,v) \ + ((sbp)->s_v1.s_reserved_for_journal = cpu_to_le16(v)) + +/* LOGGING -- */ + +/* These all interelate for performance. +** +** If the journal block count is smaller than n transactions, you lose speed. +** I don't know what n is yet, I'm guessing 8-16. +** +** typical transaction size depends on the application, how often fsync is +** called, and how many metadata blocks you dirty in a 30 second period. +** The more small files (<16k) you use, the larger your transactions will +** be. +** +** If your journal fills faster than dirty buffers get flushed to disk, it must flush them before allowing the journal +** to wrap, which slows things down. If you need high speed meta data updates, the journal should be big enough +** to prevent wrapping before dirty meta blocks get to disk. +** +** If the batch max is smaller than the transaction max, you'll waste space at the end of the journal +** because journal_end sets the next transaction to start at 0 if the next transaction has any chance of wrapping. +** +** The large the batch max age, the better the speed, and the more meta data changes you'll lose after a crash. +** +*/ + +/* don't mess with these for a while */ + /* we have a node size define somewhere in reiserfs_fs.h. -Hans */ +#define JOURNAL_BLOCK_SIZE 4096 /* BUG gotta get rid of this */ +#define JOURNAL_MAX_CNODE 1500 /* max cnodes to allocate. */ +#define JOURNAL_HASH_SIZE 8192 +#define JOURNAL_NUM_BITMAPS 5 /* number of copies of the bitmaps to have floating. Must be >= 2 */ + +/* One of these for every block in every transaction +** Each one is in two hash tables. First, a hash of the current transaction, and after journal_end, a +** hash of all the in memory transactions. +** next and prev are used by the current transaction (journal_hash). +** hnext and hprev are used by journal_list_hash. If a block is in more than one transaction, the journal_list_hash +** links it in multiple times. This allows flush_journal_list to remove just the cnode belonging +** to a given transaction. +*/ +struct reiserfs_journal_cnode { + struct buffer_head *bh; /* real buffer head */ + struct super_block *sb; /* dev of real buffer head */ + __u32 blocknr; /* block number of real buffer head, == 0 when buffer on disk */ + unsigned long state; + struct reiserfs_journal_list *jlist; /* journal list this cnode lives in */ + struct reiserfs_journal_cnode *next; /* next in transaction list */ + struct reiserfs_journal_cnode *prev; /* prev in transaction list */ + struct reiserfs_journal_cnode *hprev; /* prev in hash list */ + struct reiserfs_journal_cnode *hnext; /* next in hash list */ +}; + +struct reiserfs_bitmap_node { + int id; + char *data; + struct list_head list; +}; + +struct reiserfs_list_bitmap { + struct reiserfs_journal_list *journal_list; + struct reiserfs_bitmap_node **bitmaps; +}; + +/* +** one of these for each transaction. The most important part here is the j_realblock. +** this list of cnodes is used to hash all the blocks in all the commits, to mark all the +** real buffer heads dirty once all the commits hit the disk, +** and to make sure every real block in a transaction is on disk before allowing the log area +** to be overwritten */ +struct reiserfs_journal_list { + unsigned long j_start; + unsigned long j_state; + unsigned long j_len; + atomic_t j_nonzerolen; + atomic_t j_commit_left; + atomic_t j_older_commits_done; /* all commits older than this on disk */ + struct mutex j_commit_mutex; + unsigned int j_trans_id; + time_t j_timestamp; + struct reiserfs_list_bitmap *j_list_bitmap; + struct buffer_head *j_commit_bh; /* commit buffer head */ + struct reiserfs_journal_cnode *j_realblock; + struct reiserfs_journal_cnode *j_freedlist; /* list of buffers that were freed during this trans. free each of these on flush */ + /* time ordered list of all active transactions */ + struct list_head j_list; + + /* time ordered list of all transactions we haven't tried to flush yet */ + struct list_head j_working_list; + + /* list of tail conversion targets in need of flush before commit */ + struct list_head j_tail_bh_list; + /* list of data=ordered buffers in need of flush before commit */ + struct list_head j_bh_list; + int j_refcount; +}; + +struct reiserfs_journal { + struct buffer_head **j_ap_blocks; /* journal blocks on disk */ + struct reiserfs_journal_cnode *j_last; /* newest journal block */ + struct reiserfs_journal_cnode *j_first; /* oldest journal block. start here for traverse */ + + struct block_device *j_dev_bd; + fmode_t j_dev_mode; + int j_1st_reserved_block; /* first block on s_dev of reserved area journal */ + + unsigned long j_state; + unsigned int j_trans_id; + unsigned long j_mount_id; + unsigned long j_start; /* start of current waiting commit (index into j_ap_blocks) */ + unsigned long j_len; /* length of current waiting commit */ + unsigned long j_len_alloc; /* number of buffers requested by journal_begin() */ + atomic_t j_wcount; /* count of writers for current commit */ + unsigned long j_bcount; /* batch count. allows turning X transactions into 1 */ + unsigned long j_first_unflushed_offset; /* first unflushed transactions offset */ + unsigned j_last_flush_trans_id; /* last fully flushed journal timestamp */ + struct buffer_head *j_header_bh; + + time_t j_trans_start_time; /* time this transaction started */ + struct mutex j_mutex; + struct mutex j_flush_mutex; + wait_queue_head_t j_join_wait; /* wait for current transaction to finish before starting new one */ + atomic_t j_jlock; /* lock for j_join_wait */ + int j_list_bitmap_index; /* number of next list bitmap to use */ + int j_must_wait; /* no more journal begins allowed. MUST sleep on j_join_wait */ + int j_next_full_flush; /* next journal_end will flush all journal list */ + int j_next_async_flush; /* next journal_end will flush all async commits */ + + int j_cnode_used; /* number of cnodes on the used list */ + int j_cnode_free; /* number of cnodes on the free list */ + + unsigned int j_trans_max; /* max number of blocks in a transaction. */ + unsigned int j_max_batch; /* max number of blocks to batch into a trans */ + unsigned int j_max_commit_age; /* in seconds, how old can an async commit be */ + unsigned int j_max_trans_age; /* in seconds, how old can a transaction be */ + unsigned int j_default_max_commit_age; /* the default for the max commit age */ + + struct reiserfs_journal_cnode *j_cnode_free_list; + struct reiserfs_journal_cnode *j_cnode_free_orig; /* orig pointer returned from vmalloc */ + + struct reiserfs_journal_list *j_current_jl; + int j_free_bitmap_nodes; + int j_used_bitmap_nodes; + + int j_num_lists; /* total number of active transactions */ + int j_num_work_lists; /* number that need attention from kreiserfsd */ + + /* debugging to make sure things are flushed in order */ + unsigned int j_last_flush_id; + + /* debugging to make sure things are committed in order */ + unsigned int j_last_commit_id; + + struct list_head j_bitmap_nodes; + struct list_head j_dirty_buffers; + spinlock_t j_dirty_buffers_lock; /* protects j_dirty_buffers */ + + /* list of all active transactions */ + struct list_head j_journal_list; + /* lists that haven't been touched by writeback attempts */ + struct list_head j_working_list; + + struct reiserfs_list_bitmap j_list_bitmap[JOURNAL_NUM_BITMAPS]; /* array of bitmaps to record the deleted blocks */ + struct reiserfs_journal_cnode *j_hash_table[JOURNAL_HASH_SIZE]; /* hash table for real buffer heads in current trans */ + struct reiserfs_journal_cnode *j_list_hash_table[JOURNAL_HASH_SIZE]; /* hash table for all the real buffer heads in all + the transactions */ + struct list_head j_prealloc_list; /* list of inodes which have preallocated blocks */ + int j_persistent_trans; + unsigned long j_max_trans_size; + unsigned long j_max_batch_size; + + int j_errno; + + /* when flushing ordered buffers, throttle new ordered writers */ + struct delayed_work j_work; + struct super_block *j_work_sb; + atomic_t j_async_throttle; +}; + +enum journal_state_bits { + J_WRITERS_BLOCKED = 1, /* set when new writers not allowed */ + J_WRITERS_QUEUED, /* set when log is full due to too many writers */ + J_ABORTED, /* set when log is aborted */ +}; + +#define JOURNAL_DESC_MAGIC "ReIsErLB" /* ick. magic string to find desc blocks in the journal */ + +typedef __u32(*hashf_t) (const signed char *, int); + +struct reiserfs_bitmap_info { + __u32 free_count; +}; + +struct proc_dir_entry; + +#if defined( CONFIG_PROC_FS ) && defined( CONFIG_REISERFS_PROC_INFO ) +typedef unsigned long int stat_cnt_t; +typedef struct reiserfs_proc_info_data { + spinlock_t lock; + int exiting; + int max_hash_collisions; + + stat_cnt_t breads; + stat_cnt_t bread_miss; + stat_cnt_t search_by_key; + stat_cnt_t search_by_key_fs_changed; + stat_cnt_t search_by_key_restarted; + + stat_cnt_t insert_item_restarted; + stat_cnt_t paste_into_item_restarted; + stat_cnt_t cut_from_item_restarted; + stat_cnt_t delete_solid_item_restarted; + stat_cnt_t delete_item_restarted; + + stat_cnt_t leaked_oid; + stat_cnt_t leaves_removable; + + /* balances per level. Use explicit 5 as MAX_HEIGHT is not visible yet. */ + stat_cnt_t balance_at[5]; /* XXX */ + /* sbk == search_by_key */ + stat_cnt_t sbk_read_at[5]; /* XXX */ + stat_cnt_t sbk_fs_changed[5]; + stat_cnt_t sbk_restarted[5]; + stat_cnt_t items_at[5]; /* XXX */ + stat_cnt_t free_at[5]; /* XXX */ + stat_cnt_t can_node_be_removed[5]; /* XXX */ + long int lnum[5]; /* XXX */ + long int rnum[5]; /* XXX */ + long int lbytes[5]; /* XXX */ + long int rbytes[5]; /* XXX */ + stat_cnt_t get_neighbors[5]; + stat_cnt_t get_neighbors_restart[5]; + stat_cnt_t need_l_neighbor[5]; + stat_cnt_t need_r_neighbor[5]; + + stat_cnt_t free_block; + struct __scan_bitmap_stats { + stat_cnt_t call; + stat_cnt_t wait; + stat_cnt_t bmap; + stat_cnt_t retry; + stat_cnt_t in_journal_hint; + stat_cnt_t in_journal_nohint; + stat_cnt_t stolen; + } scan_bitmap; + struct __journal_stats { + stat_cnt_t in_journal; + stat_cnt_t in_journal_bitmap; + stat_cnt_t in_journal_reusable; + stat_cnt_t lock_journal; + stat_cnt_t lock_journal_wait; + stat_cnt_t journal_being; + stat_cnt_t journal_relock_writers; + stat_cnt_t journal_relock_wcount; + stat_cnt_t mark_dirty; + stat_cnt_t mark_dirty_already; + stat_cnt_t mark_dirty_notjournal; + stat_cnt_t restore_prepared; + stat_cnt_t prepare; + stat_cnt_t prepare_retry; + } journal; +} reiserfs_proc_info_data_t; +#else +typedef struct reiserfs_proc_info_data { +} reiserfs_proc_info_data_t; +#endif + +/* reiserfs union of in-core super block data */ +struct reiserfs_sb_info { + struct buffer_head *s_sbh; /* Buffer containing the super block */ + /* both the comment and the choice of + name are unclear for s_rs -Hans */ + struct reiserfs_super_block *s_rs; /* Pointer to the super block in the buffer */ + struct reiserfs_bitmap_info *s_ap_bitmap; + struct reiserfs_journal *s_journal; /* pointer to journal information */ + unsigned short s_mount_state; /* reiserfs state (valid, invalid) */ + + /* Serialize writers access, replace the old bkl */ + struct mutex lock; + /* Owner of the lock (can be recursive) */ + struct task_struct *lock_owner; + /* Depth of the lock, start from -1 like the bkl */ + int lock_depth; + + /* Comment? -Hans */ + void (*end_io_handler) (struct buffer_head *, int); + hashf_t s_hash_function; /* pointer to function which is used + to sort names in directory. Set on + mount */ + unsigned long s_mount_opt; /* reiserfs's mount options are set + here (currently - NOTAIL, NOLOG, + REPLAYONLY) */ + + struct { /* This is a structure that describes block allocator options */ + unsigned long bits; /* Bitfield for enable/disable kind of options */ + unsigned long large_file_size; /* size started from which we consider file to be a large one(in blocks) */ + int border; /* percentage of disk, border takes */ + int preallocmin; /* Minimal file size (in blocks) starting from which we do preallocations */ + int preallocsize; /* Number of blocks we try to prealloc when file + reaches preallocmin size (in blocks) or + prealloc_list is empty. */ + } s_alloc_options; + + /* Comment? -Hans */ + wait_queue_head_t s_wait; + /* To be obsoleted soon by per buffer seals.. -Hans */ + atomic_t s_generation_counter; // increased by one every time the + // tree gets re-balanced + unsigned long s_properties; /* File system properties. Currently holds + on-disk FS format */ + + /* session statistics */ + int s_disk_reads; + int s_disk_writes; + int s_fix_nodes; + int s_do_balance; + int s_unneeded_left_neighbor; + int s_good_search_by_key_reada; + int s_bmaps; + int s_bmaps_without_search; + int s_direct2indirect; + int s_indirect2direct; + /* set up when it's ok for reiserfs_read_inode2() to read from + disk inode with nlink==0. Currently this is only used during + finish_unfinished() processing at mount time */ + int s_is_unlinked_ok; + reiserfs_proc_info_data_t s_proc_info_data; + struct proc_dir_entry *procdir; + int reserved_blocks; /* amount of blocks reserved for further allocations */ + spinlock_t bitmap_lock; /* this lock on now only used to protect reserved_blocks variable */ + struct dentry *priv_root; /* root of /.reiserfs_priv */ + struct dentry *xattr_root; /* root of /.reiserfs_priv/xattrs */ + int j_errno; +#ifdef CONFIG_QUOTA + char *s_qf_names[MAXQUOTAS]; + int s_jquota_fmt; +#endif + char *s_jdev; /* Stored jdev for mount option showing */ +#ifdef CONFIG_REISERFS_CHECK + + struct tree_balance *cur_tb; /* + * Detects whether more than one + * copy of tb exists per superblock + * as a means of checking whether + * do_balance is executing concurrently + * against another tree reader/writer + * on a same mount point. + */ +#endif +}; + +/* Definitions of reiserfs on-disk properties: */ +#define REISERFS_3_5 0 +#define REISERFS_3_6 1 +#define REISERFS_OLD_FORMAT 2 + +enum reiserfs_mount_options { +/* Mount options */ + REISERFS_LARGETAIL, /* large tails will be created in a session */ + REISERFS_SMALLTAIL, /* small (for files less than block size) tails will be created in a session */ + REPLAYONLY, /* replay journal and return 0. Use by fsck */ + REISERFS_CONVERT, /* -o conv: causes conversion of old + format super block to the new + format. If not specified - old + partition will be dealt with in a + manner of 3.5.x */ + +/* -o hash={tea, rupasov, r5, detect} is meant for properly mounting +** reiserfs disks from 3.5.19 or earlier. 99% of the time, this option +** is not required. If the normal autodection code can't determine which +** hash to use (because both hashes had the same value for a file) +** use this option to force a specific hash. It won't allow you to override +** the existing hash on the FS, so if you have a tea hash disk, and mount +** with -o hash=rupasov, the mount will fail. +*/ + FORCE_TEA_HASH, /* try to force tea hash on mount */ + FORCE_RUPASOV_HASH, /* try to force rupasov hash on mount */ + FORCE_R5_HASH, /* try to force rupasov hash on mount */ + FORCE_HASH_DETECT, /* try to detect hash function on mount */ + + REISERFS_DATA_LOG, + REISERFS_DATA_ORDERED, + REISERFS_DATA_WRITEBACK, + +/* used for testing experimental features, makes benchmarking new + features with and without more convenient, should never be used by + users in any code shipped to users (ideally) */ + + REISERFS_NO_BORDER, + REISERFS_NO_UNHASHED_RELOCATION, + REISERFS_HASHED_RELOCATION, + REISERFS_ATTRS, + REISERFS_XATTRS_USER, + REISERFS_POSIXACL, + REISERFS_EXPOSE_PRIVROOT, + REISERFS_BARRIER_NONE, + REISERFS_BARRIER_FLUSH, + + /* Actions on error */ + REISERFS_ERROR_PANIC, + REISERFS_ERROR_RO, + REISERFS_ERROR_CONTINUE, + + REISERFS_USRQUOTA, /* User quota option specified */ + REISERFS_GRPQUOTA, /* Group quota option specified */ + + REISERFS_TEST1, + REISERFS_TEST2, + REISERFS_TEST3, + REISERFS_TEST4, + REISERFS_UNSUPPORTED_OPT, +}; + +#define reiserfs_r5_hash(s) (REISERFS_SB(s)->s_mount_opt & (1 << FORCE_R5_HASH)) +#define reiserfs_rupasov_hash(s) (REISERFS_SB(s)->s_mount_opt & (1 << FORCE_RUPASOV_HASH)) +#define reiserfs_tea_hash(s) (REISERFS_SB(s)->s_mount_opt & (1 << FORCE_TEA_HASH)) +#define reiserfs_hash_detect(s) (REISERFS_SB(s)->s_mount_opt & (1 << FORCE_HASH_DETECT)) +#define reiserfs_no_border(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_NO_BORDER)) +#define reiserfs_no_unhashed_relocation(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_NO_UNHASHED_RELOCATION)) +#define reiserfs_hashed_relocation(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_HASHED_RELOCATION)) +#define reiserfs_test4(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_TEST4)) + +#define have_large_tails(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_LARGETAIL)) +#define have_small_tails(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_SMALLTAIL)) +#define replay_only(s) (REISERFS_SB(s)->s_mount_opt & (1 << REPLAYONLY)) +#define reiserfs_attrs(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_ATTRS)) +#define old_format_only(s) (REISERFS_SB(s)->s_properties & (1 << REISERFS_3_5)) +#define convert_reiserfs(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_CONVERT)) +#define reiserfs_data_log(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_DATA_LOG)) +#define reiserfs_data_ordered(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_DATA_ORDERED)) +#define reiserfs_data_writeback(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_DATA_WRITEBACK)) +#define reiserfs_xattrs_user(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_XATTRS_USER)) +#define reiserfs_posixacl(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_POSIXACL)) +#define reiserfs_expose_privroot(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_EXPOSE_PRIVROOT)) +#define reiserfs_xattrs_optional(s) (reiserfs_xattrs_user(s) || reiserfs_posixacl(s)) +#define reiserfs_barrier_none(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_BARRIER_NONE)) +#define reiserfs_barrier_flush(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_BARRIER_FLUSH)) + +#define reiserfs_error_panic(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_ERROR_PANIC)) +#define reiserfs_error_ro(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_ERROR_RO)) + +void reiserfs_file_buffer(struct buffer_head *bh, int list); +extern struct file_system_type reiserfs_fs_type; +int reiserfs_resize(struct super_block *, unsigned long); + +#define CARRY_ON 0 +#define SCHEDULE_OCCURRED 1 + +#define SB_BUFFER_WITH_SB(s) (REISERFS_SB(s)->s_sbh) +#define SB_JOURNAL(s) (REISERFS_SB(s)->s_journal) +#define SB_JOURNAL_1st_RESERVED_BLOCK(s) (SB_JOURNAL(s)->j_1st_reserved_block) +#define SB_JOURNAL_LEN_FREE(s) (SB_JOURNAL(s)->j_journal_len_free) +#define SB_AP_BITMAP(s) (REISERFS_SB(s)->s_ap_bitmap) + +#define SB_DISK_JOURNAL_HEAD(s) (SB_JOURNAL(s)->j_header_bh->) + +/* A safe version of the "bdevname", which returns the "s_id" field of + * a superblock or else "Null superblock" if the super block is NULL. + */ +static inline char *reiserfs_bdevname(struct super_block *s) +{ + return (s == NULL) ? "Null superblock" : s->s_id; +} + +#define reiserfs_is_journal_aborted(journal) (unlikely (__reiserfs_is_journal_aborted (journal))) +static inline int __reiserfs_is_journal_aborted(struct reiserfs_journal + *journal) +{ + return test_bit(J_ABORTED, &journal->j_state); +} + +/* + * 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)) */ < |