reiserfs.h 117 KB
Newer Older
1
/*
2 3
 * Copyright 1996, 1997, 1998 Hans Reiser, see reiserfs/README for
 * licensing and copyright details
4 5 6 7 8 9 10
 */

#include <linux/reiserfs_fs.h>

#include <linux/slab.h>
#include <linux/interrupt.h>
#include <linux/sched.h>
11
#include <linux/bug.h>
12 13 14 15 16 17 18 19 20 21 22 23 24
#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

Al Viro's avatar
Al Viro committed
25 26
struct reiserfs_journal_list;

27
/* bitmasks for i_flags field in reiserfs-specific part of inode */
Al Viro's avatar
Al Viro committed
28
typedef enum {
29 30 31 32
	/*
	 * 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
	 */
Al Viro's avatar
Al Viro committed
33
	i_item_key_version_mask = 0x0001,
34 35 36 37 38

	/*
	 * If this is unset, object has 3.5 stat data, otherwise,
	 * it has 3.6 stat data with 64bit size, 32bit nlink etc.
	 */
Al Viro's avatar
Al Viro committed
39
	i_stat_data_version_mask = 0x0002,
40 41

	/* file might need tail packing on close */
Al Viro's avatar
Al Viro committed
42
	i_pack_on_close_mask = 0x0004,
43 44

	/* don't pack tail of file */
Al Viro's avatar
Al Viro committed
45
	i_nopack_mask = 0x0008,
46 47 48 49 50 51

	/*
	 * If either of these are 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.
	 */
Al Viro's avatar
Al Viro committed
52 53
	i_link_saved_unlink_mask = 0x0010,
	i_link_saved_truncate_mask = 0x0020,
54

Al Viro's avatar
Al Viro committed
55 56 57 58 59 60
	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 */
61 62 63 64 65

	/*
	 * transient inode flags that are never stored on disk. Bitmasks
	 * for this field are defined above.
	 */
Al Viro's avatar
Al Viro committed
66 67
	__u32 i_flags;

68 69
	/* offset of first byte stored in direct item. */
	__u32 i_first_direct_byte;
Al Viro's avatar
Al Viro committed
70 71 72 73

	/* copy of persistent inode flags read from sd_attrs. */
	__u32 i_attrs;

74 75
	/* first unused block of a sequence of unused blocks */
	int i_prealloc_block;
Al Viro's avatar
Al Viro committed
76 77
	int i_prealloc_count;	/* length of that sequence */

78 79 80 81 82 83 84 85
	/* per-transaction list of inodes which  have preallocated blocks */
	struct list_head i_prealloc_list;

	/*
	 * new_packing_locality is created; new blocks for the contents
	 * of this directory should be displaced
	 */
	unsigned new_packing_locality:1;
Al Viro's avatar
Al Viro committed
86

87 88 89 90 91
	/*
	 * 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
	 */
Al Viro's avatar
Al Viro committed
92
	unsigned int i_trans_id;
93

Al Viro's avatar
Al Viro committed
94 95 96 97 98 99 100 101 102 103 104 105 106
	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;

107 108 109 110
/*
 * struct reiserfs_super_block accessors/mutators since this is a disk
 * structure, it will always be in little endian format.
 */
Al Viro's avatar
Al Viro committed
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
#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 -- */

178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
/*
 * 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.
 */
Al Viro's avatar
Al Viro committed
201 202

/* don't mess with these for a while */
203
/* we have a node size define somewhere in reiserfs_fs.h. -Hans */
Al Viro's avatar
Al Viro committed
204 205 206
#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
207 208 209 210 211 212 213 214 215 216 217 218 219 220

/* number of copies of the bitmaps to have floating.  Must be >= 2 */
#define JOURNAL_NUM_BITMAPS 5

/*
 * 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.
 */
Al Viro's avatar
Al Viro committed
221 222 223
struct reiserfs_journal_cnode {
	struct buffer_head *bh;	/* real buffer head */
	struct super_block *sb;	/* dev of real buffer head */
224 225 226 227

	/* block number of real buffer head, == 0 when buffer on disk */
	__u32 blocknr;

Al Viro's avatar
Al Viro committed
228
	unsigned long state;
229 230 231 232

	/* journal list this cnode lives in */
	struct reiserfs_journal_list *jlist;

Al Viro's avatar
Al Viro committed
233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250
	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;
};

/*
251 252 253 254 255 256
 * 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
 */
Al Viro's avatar
Al Viro committed
257 258 259 260 261 262
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;
263 264 265 266

	/* all commits older than this on disk */
	atomic_t j_older_commits_done;

Al Viro's avatar
Al Viro committed
267 268 269 270 271 272 273 274 275 276
	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;

277 278 279 280
	/*
	 * time ordered list of all transactions we haven't tried
	 * to flush yet
	 */
Al Viro's avatar
Al Viro committed
281 282 283 284
	struct list_head j_working_list;

	/* list of tail conversion targets in need of flush before commit */
	struct list_head j_tail_bh_list;
285

Al Viro's avatar
Al Viro committed
286 287 288 289 290 291 292
	/* 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 */
293 294 295 296 297
	/* newest journal block */
	struct reiserfs_journal_cnode *j_last;

	/* oldest journal block.  start here for traverse */
	struct reiserfs_journal_cnode *j_first;
Al Viro's avatar
Al Viro committed
298 299 300

	struct block_device *j_dev_bd;
	fmode_t j_dev_mode;
301 302 303

	/* first block on s_dev of reserved area journal */
	int j_1st_reserved_block;
Al Viro's avatar
Al Viro committed
304 305 306 307

	unsigned long j_state;
	unsigned int j_trans_id;
	unsigned long j_mount_id;
308 309 310

	/* start of current waiting commit (index into j_ap_blocks) */
	unsigned long j_start;
Al Viro's avatar
Al Viro committed
311
	unsigned long j_len;	/* length of current waiting commit */
312 313 314 315

	/* number of buffers requested by journal_begin() */
	unsigned long j_len_alloc;

Al Viro's avatar
Al Viro committed
316
	atomic_t j_wcount;	/* count of writers for current commit */
317 318 319 320 321 322 323 324 325 326

	/* batch count. allows turning X transactions into 1 */
	unsigned long j_bcount;

	/* first unflushed transactions offset */
	unsigned long j_first_unflushed_offset;

	/* last fully flushed journal timestamp */
	unsigned j_last_flush_trans_id;

Al Viro's avatar
Al Viro committed
327 328 329 330 331
	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;
332 333 334 335 336

	/* wait for current transaction to finish before starting new one */
	wait_queue_head_t j_join_wait;

	atomic_t j_jlock;		/* lock for j_join_wait */
Al Viro's avatar
Al Viro committed
337
	int j_list_bitmap_index;	/* number of next list bitmap to use */
338 339 340 341 342 343 344 345 346

	/* no more journal begins allowed. MUST sleep on j_join_wait */
	int j_must_wait;

	/* next journal_end will flush all journal list */
	int j_next_full_flush;

	/* next journal_end will flush all async commits */
	int j_next_async_flush;
Al Viro's avatar
Al Viro committed
347 348 349 350

	int j_cnode_used;	/* number of cnodes on the used list */
	int j_cnode_free;	/* number of cnodes on the free list */

351 352 353 354 355 356 357 358 359 360 361 362 363 364
	/* max number of blocks in a transaction.  */
	unsigned int j_trans_max;

	/* max number of blocks to batch into a trans */
	unsigned int j_max_batch;

	/* in seconds, how old can an async commit be */
	unsigned int j_max_commit_age;

	/* in seconds, how old can a transaction be */
	unsigned int j_max_trans_age;

	/* the default for the max commit age */
	unsigned int j_default_max_commit_age;
Al Viro's avatar
Al Viro committed
365 366

	struct reiserfs_journal_cnode *j_cnode_free_list;
367 368 369

	/* orig pointer returned from vmalloc */
	struct reiserfs_journal_cnode *j_cnode_free_orig;
Al Viro's avatar
Al Viro committed
370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389

	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;
390

Al Viro's avatar
Al Viro committed
391 392 393
	/* lists that haven't been touched by writeback attempts */
	struct list_head j_working_list;

394 395 396 397 398 399 400 401 402 403 404
	/* hash table for real buffer heads in current trans */
	struct reiserfs_journal_cnode *j_hash_table[JOURNAL_HASH_SIZE];

	/* hash table for all the real buffer heads in all the transactions */
	struct reiserfs_journal_cnode *j_list_hash_table[JOURNAL_HASH_SIZE];

	/* array of bitmaps to record the deleted blocks */
	struct reiserfs_list_bitmap j_list_bitmap[JOURNAL_NUM_BITMAPS];

	/* list of inodes which have preallocated blocks */
	struct list_head j_prealloc_list;
Al Viro's avatar
Al Viro committed
405 406 407 408 409 410 411 412 413 414 415 416 417 418
	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 */
419 420
	J_WRITERS_QUEUED,    /* set when log is full due to too many writers */
	J_ABORTED,           /* set when log is aborted */
Al Viro's avatar
Al Viro committed
421 422
};

423 424
/* ick.  magic string to find desc blocks in the journal */
#define JOURNAL_DESC_MAGIC "ReIsErLB"
Al Viro's avatar
Al Viro committed
425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455

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;

456 457 458 459
	/*
	 * balances per level.
	 * Use explicit 5 as MAX_HEIGHT is not visible yet.
	 */
Al Viro's avatar
Al Viro committed
460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
	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 {
511 512 513 514 515
	/* Buffer containing the super block */
	struct buffer_head *s_sbh;

	/* Pointer to the on-disk super block in the buffer */
	struct reiserfs_super_block *s_rs;
Al Viro's avatar
Al Viro committed
516
	struct reiserfs_bitmap_info *s_ap_bitmap;
517 518 519 520

	/* pointer to journal information */
	struct reiserfs_journal *s_journal;

Al Viro's avatar
Al Viro committed
521 522 523 524
	unsigned short s_mount_state;	/* reiserfs state (valid, invalid) */

	/* Serialize writers access, replace the old bkl */
	struct mutex lock;
525

Al Viro's avatar
Al Viro committed
526 527
	/* Owner of the lock (can be recursive) */
	struct task_struct *lock_owner;
528

Al Viro's avatar
Al Viro committed
529 530 531
	/* Depth of the lock, start from -1 like the bkl */
	int lock_depth;

532 533
	struct workqueue_struct *commit_wq;

Al Viro's avatar
Al Viro committed
534 535
	/* Comment? -Hans */
	void (*end_io_handler) (struct buffer_head *, int);
536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556

	/*
	 * pointer to function which is used to sort names in directory.
	 * Set on mount
	 */
	hashf_t s_hash_function;

	/* reiserfs's mount options are set here */
	unsigned long s_mount_opt;

	/* This is a structure that describes block allocator options */
	struct {
		/* Bitfield for enable/disable kind of options */
		unsigned long bits;

		/*
		 * size started from which we consider file
		 * to be a large one (in blocks)
		 */
		unsigned long large_file_size;

Al Viro's avatar
Al Viro committed
557
		int border;	/* percentage of disk, border takes */
558 559 560 561 562 563 564 565 566 567 568 569 570

		/*
		 * Minimal file size (in blocks) starting
		 * from which we do preallocations
		 */
		int preallocmin;

		/*
		 * Number of blocks we try to prealloc when file
		 * reaches preallocmin size (in blocks) or prealloc_list
		 is empty.
		 */
		int preallocsize;
Al Viro's avatar
Al Viro committed
571 572 573 574
	} s_alloc_options;

	/* Comment? -Hans */
	wait_queue_head_t s_wait;
575 576 577 578 579
	/* increased by one every time the  tree gets re-balanced */
	atomic_t s_generation_counter;

	/* File system properties. Currently holds on-disk FS format */
	unsigned long s_properties;
Al Viro's avatar
Al Viro committed
580 581 582 583 584 585 586 587 588 589 590 591

	/* 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;
592 593 594 595 596 597

	/*
	 * 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
	 */
Al Viro's avatar
Al Viro committed
598
	int s_is_unlinked_ok;
599

Al Viro's avatar
Al Viro committed
600 601
	reiserfs_proc_info_data_t s_proc_info_data;
	struct proc_dir_entry *procdir;
602 603 604 605 606 607 608

	/* amount of blocks reserved for further allocations */
	int reserved_blocks;


	/* this lock on now only used to protect reserved_blocks variable */
	spinlock_t bitmap_lock;
Al Viro's avatar
Al Viro committed
609 610 611
	struct dentry *priv_root;	/* root of /.reiserfs_priv */
	struct dentry *xattr_root;	/* root of /.reiserfs_priv/xattrs */
	int j_errno;
612 613 614 615 616

	int work_queued;              /* non-zero delayed work is queued */
	struct delayed_work old_work; /* old transactions flush delayed work */
	spinlock_t old_work_lock;     /* protects old_work and work_queued */

Al Viro's avatar
Al Viro committed
617 618 619 620 621 622 623
#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

624 625 626 627 628 629 630
	/*
	 * 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.
	 */
	struct tree_balance *cur_tb;
Al Viro's avatar
Al Viro committed
631 632 633 634 635 636 637 638 639
#endif
};

/* Definitions of reiserfs on-disk properties: */
#define REISERFS_3_5 0
#define REISERFS_3_6 1
#define REISERFS_OLD_FORMAT 2

/* Mount options */
640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668
enum reiserfs_mount_options {
	/* large tails will be created in a session */
	REISERFS_LARGETAIL,
	/*
	 * small (for files less than block size) tails will
	 * be created in a session
	 */
	REISERFS_SMALLTAIL,

	/* replay journal and return 0. Use by fsck */
	REPLAYONLY,

	/*
	 * -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
	 */
	REISERFS_CONVERT,

	/*
	 * -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.
	 */
Al Viro's avatar
Al Viro committed
669 670 671 672 673 674 675 676 677
	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,

678 679 680 681 682
	/*
	 * 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)
	 */
Al Viro's avatar
Al Viro committed
683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758

	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->)

#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);
}

759 760 761 762 763 764 765
/*
 * 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);
766 767
int __must_check reiserfs_write_unlock_nested(struct super_block *s);
void reiserfs_write_lock_nested(struct super_block *s, int depth);
768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802

#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,
803
					    struct super_block *s)
804
{
805 806 807
	int depth;

	depth = reiserfs_write_unlock_nested(s);
808
	mutex_lock(m);
809
	reiserfs_write_lock_nested(s, depth);
810 811 812 813
}

static inline void
reiserfs_mutex_lock_nested_safe(struct mutex *m, unsigned int subclass,
814
				struct super_block *s)
815
{
816 817 818
	int depth;

	depth = reiserfs_write_unlock_nested(s);
819
	mutex_lock_nested(m, subclass);
820
	reiserfs_write_lock_nested(s, depth);
821 822 823 824 825
}

static inline void
reiserfs_down_read_safe(struct rw_semaphore *sem, struct super_block *s)
{
826 827 828 829
       int depth;
       depth = reiserfs_write_unlock_nested(s);
       down_read(sem);
       reiserfs_write_lock_nested(s, depth);
830 831 832 833 834 835 836 837 838
}

/*
 * 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()) {
839 840 841
		int depth;

		depth = reiserfs_write_unlock_nested(s);
842
		schedule();
843
		reiserfs_write_lock_nested(s, depth);
844 845 846 847 848
	}
}

struct fid;

849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
/*
 * 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.
 */
871 872 873 874 875 876 877 878 879 880

#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))

881 882 883 884 885
/*
 * 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
 */
886 887
#define ROUND_UP(x) _ROUND_UP(x,8LL)

888 889 890 891
/*
 * debug levels.  Right now, CONFIG_REISERFS_CHECK means print all debug
 * messages.
 */
892 893 894 895 896 897 898 899
#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 */

900
/* always check a condition and panic if it's false. */
901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922
#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
 */

923 924 925
/***************************************************************************
 *                             SUPER BLOCK                                 *
 ***************************************************************************/
926 927

/*
928 929 930
 * 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.
931
 */
932
#define UNSET_HASH 0	/* Detect hash on disk */
933 934 935 936 937 938
#define TEA_HASH  1
#define YURA_HASH 2
#define R5_HASH   3
#define DEFAULT_HASH R5_HASH

struct journal_params {
939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964
	/* where does journal start from on its * device */
	__le32 jp_journal_1st_block;

	/* journal device st_rdev */
	__le32 jp_journal_dev;

	/* size of the journal */
	__le32 jp_journal_size;

	/* max number of blocks in a transaction. */
	__le32 jp_journal_trans_max;

	/*
	 * random value made on fs creation
	 * (this was sb_journal_block_count)
	 */
	__le32 jp_journal_magic;

	/* max number of blocks to batch into a trans */
	__le32 jp_journal_max_batch;

	/* in seconds, how old can an async  commit be */
	__le32 jp_journal_max_commit_age;

	/* in seconds, how old can a transaction be */
	__le32 jp_journal_max_trans_age;
965 966 967 968 969 970 971 972 973
};

/* 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 */
974 975 976

	/* max size of object id array, see get_objectid() commentary  */
	__le16 s_oid_maxsize;
977
	__le16 s_oid_cursize;	/* current size of object id array */
978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997

	/* this is set to 1 when filesystem was umounted, to 2 - when not */
	__le16 s_umount_state;

	/*
	 * reiserfs magic string indicates that file system is reiserfs:
	 * "ReIsErFs" or "ReIsEr2Fs" or "ReIsEr3Fs"
	 */
	char s_magic[10];

	/*
	 * it is set to used by fsck to mark which
	 * phase of rebuilding is done
	 */
	__le16 s_fs_state;
	/*
	 * indicate, what hash function is being use
	 * to sort names in a directory
	 */
	__le32 s_hash_function_code;
998
	__le16 s_tree_height;	/* height of disk tree */
999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015

	/*
	 * amount of bitmap blocks needed to address
	 * each block of file system
	 */
	__le16 s_bmap_nr;

	/*
	 * this field is only reliable on filesystem with non-standard journal
	 */
	__le16 s_version;

	/*
	 * size in blocks of journal area on main device, we need to
	 * keep after making fs with non-standard journal
	 */
	__le16 s_reserved_for_journal;
1016 1017 1018 1019 1020 1021 1022 1023
} __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;
1024 1025 1026 1027

	/* Right now used only by inode-attributes, if enabled */
	__le32 s_flags;

1028 1029 1030 1031 1032 1033
	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 */
1034 1035 1036 1037 1038

	/*
	 * zero filled by mkreiserfs and reiserfs_convert_objectid_map_v1()
	 * so any additions must be updated there as well. */
	char s_unused[76];
1039 1040 1041 1042 1043 1044 1045
} __attribute__ ((__packed__));

#define SB_SIZE (sizeof(struct reiserfs_super_block))

#define REISERFS_VERSION_1 0
#define REISERFS_VERSION_2 2

1046
/* on-disk super block fields converted to cpu form */
1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100
#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);

1101 1102 1103 1104 1105 1106 1107
/*
 * 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
 */
1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131
#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;
};

1132
/* there are two formats of keys: 3.5 and 3.6 */
1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
#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;
}

1150 1151 1152 1153
/*
 * Don't trust REISERFS_SB(sb)->s_bmap_nr, it's a u16
 * which overflows on large file systems.
 */
1154 1155 1156 1157 1158 1159 1160 1161 1162 1163
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);
}

1164 1165 1166 1167
/*
 * this says about version of key of all items (but stat data) the
 * object consists of
 */
1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185
#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; })

1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197
/*
 * 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
 */
1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210
#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) ) ) \
)

1211 1212 1213 1214 1215
/*
 * 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.
1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228
*/
#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

1229 1230 1231
/*
 * there are 5 item types currently
 */
1232 1233 1234 1235 1236
#define TYPE_STAT_DATA 0
#define TYPE_INDIRECT 1
#define TYPE_DIRECT 2
#define TYPE_DIRENTRY 3
#define TYPE_MAXTYPE 3
1237
#define TYPE_ANY 15		/* FIXME: comment is required */
1238

1239 1240 1241
/***************************************************************************
 *                       KEY & ITEM HEAD                                   *
 ***************************************************************************/
1242

1243
/* * directories use this key as well as old files */
1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275
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);
}

1276 1277 1278 1279
/*
 * Key of an item determines its location in the S+tree, and
 * is composed of 4 components
 */
1280
struct reiserfs_key {
1281 1282 1283
	/* packing locality: by default parent directory object id */
	__le32 k_dir_id;

1284 1285 1286 1287 1288 1289 1290 1291
	__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 {
1292 1293
	/* packing locality: by default parent directory object id */
	__u32 k_dir_id;
1294 1295 1296 1297 1298 1299 1300 1301
	__u32 k_objectid;	/* object identifier */
	__u64 k_offset;
	__u8 k_type;
};

struct cpu_key {
	struct in_core_key on_disk_key;
	int version;
1302 1303
	/* 3 in all cases but direct2indirect and indirect2direct conversion */
	int key_length;
1304 1305
};

1306 1307 1308 1309 1310 1311
/*
 * 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.
 */
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339
#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

1340
/* return values for reiserfs_find_entry and search_by_entry_key */
1341 1342 1343 1344 1345
#define NAME_FOUND 1
#define NAME_NOT_FOUND 0
#define GOTO_PREVIOUS_ITEM 2
#define NAME_FOUND_INVISIBLE 3

1346 1347 1348 1349 1350 1351
/*
 * 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.
 */
1352 1353

struct item_head {
1354 1355 1356 1357
	/*
	 * Everything in the tree is found by searching for it based on
	 * its key.
	 */
1358 1359
	struct reiserfs_key ih_key;
	union {
1360 1361 1362 1363 1364 1365 1366 1367
		/*
		 * 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.
		 */
1368
		__le16 ih_free_space_reserved;
1369 1370 1371 1372 1373

		/*
		 * Iff this is a directory item, this field equals the
		 * number of directory entries in the directory item.
		 */
1374 1375 1376
		__le16 ih_entry_count;
	} __attribute__ ((__packed__)) u;
	__le16 ih_item_len;	/* total size of the item body */
1377 1378 1379 1380 1381 1382 1383 1384 1385

	/* an offset to the item body within the block */
	__le16 ih_item_location;

	/*
	 * 0 for all old items, 2 for new ones. Highest bit is set by fsck
	 * temporary, cleaned after all done
	 */
	__le16 ih_version;
1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
} __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)))

1407 1408 1409 1410 1411 1412 1413
/*
 * 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.
 */
1414 1415 1416
#define get_block_num(p, i) get_unaligned_le32((p) + (i))
#define put_block_num(p, i, v) put_unaligned_le32((v), (p) + (i))

1417
/* * in old version uniqueness field shows key type */
1418 1419 1420 1421
#define V1_SD_UNIQUENESS 0
#define V1_INDIRECT_UNIQUENESS 0xfffffffe
#define V1_DIRECT_UNIQUENESS 0xffffffff
#define V1_DIRENTRY_UNIQUENESS 500
1422
#define V1_ANY_UNIQUENESS 555	/* FIXME: comment is required */
1423

1424
/* here are conversion routines */
1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460
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;
	}
}

1461 1462 1463 1464 1465
/*
 * 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
 */
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480
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)
{
1481 1482 1483 1484 1485
	if (version == KEY_FORMAT_3_5) {
		loff_t val = le32_to_cpu(key->u.k_offset_v1.k_uniqueness);
		return uniqueness2type(val);
	} else
		return offset_v2_k_type(&(key->u.k_offset_v2));
1486 1487 1488 1489 1490 1491 1492 1493 1494 1495
}

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)
{
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511
	if (version == KEY_FORMAT_3_5)
		key->u.k_offset_v1.k_offset = cpu_to_le32(offset);
	else
		set_offset_v2_k_offset(&key->u.k_offset_v2, offset);
}

static inline void add_le_key_k_offset(int version, struct reiserfs_key *key,
				       loff_t offset)
{
	set_le_key_k_offset(version, key,
			    le_key_k_offset(version, key) + offset);
}

static inline void add_le_ih_k_offset(struct item_head *ih, loff_t offset)
{
	add_le_key_k_offset(ih_version(ih), &(ih->ih_key), offset);
1512 1513 1514 1515 1516 1517 1518 1519 1520 1521
}

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)
{
1522 1523 1524 1525 1526
	if (version == KEY_FORMAT_3_5) {
		type = type2uniqueness(type);
		key->u.k_offset_v1.k_uniqueness = cpu_to_le32(type);
	} else
	       set_offset_v2_k_type(&key->u.k_offset_v2, type);
1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553
}

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;
}

1554
/* item header has version.  */
1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574
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);
}

1575
/* key is pointer to cpu key, result is cpu */
1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625
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;

1626
/*
1627 1628 1629 1630 1631 1632 1633 1634
 * 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   |
 * |______|_______________|___________________|___________|
 */

1635 1636 1637 1638
/*
 * Header of a disk block.  More precisely, header of a formatted leaf
 * or internal node, and not the header of an unformatted node.
 */
1639 1640 1641 1642 1643 1644
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 */
1645 1646 1647

	/* kept only for compatibility */
	struct reiserfs_key blk_right_delim_key;
1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661
};

#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)

1662 1663
/* values for blk_level field of the struct block_head */

1664
/*
1665 1666
 * 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
1667
 */
1668
#define FREE_LEVEL 0
1669 1670 1671

#define DISK_LEAF_NODE_LEVEL  1	/* Leaf node level. */

1672 1673 1674 1675
/*
 * Given the buffer head of a formatted node, resolve to the
 * block head of that node.
 */
1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695
#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)

1696 1697 1698
/***************************************************************************
 *                             STAT DATA                                   *
 ***************************************************************************/
1699

1700 1701 1702 1703
/*
 * old stat data is 32 bytes long. We are going to distinguish new one by
 * different size
*/
1704 1705 1706 1707 1708 1709 1710 1711
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  */
1712 1713 1714 1715 1716 1717

	/*
	 * time inode (stat data) was last changed
	 * (except changes to sd_atime and sd_mtime)
	 */
	__le32 sd_ctime;
1718 1719 1720 1721
	union {
		__le32 sd_rdev;
		__le32 sd_blocks;	/* number of blocks file uses */
	} __attribute__ ((__packed__)) u;
1722 1723 1724 1725 1726 1727 1728 1729 1730

	/*
	 * 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
	 */
	__le32 sd_first_direct_byte;
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761
} __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) */

1762 1763 1764 1765
/*
 * we want common flags to have the same values as in ext2,
 * so chattr(1) will work without problems
 */
1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
#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 )

1785 1786 1787 1788
/*
 * Stat Data on disk (reiserfs version of UFS disk inode minus the
 * address blocks)
 */
1789 1790 1791 1792 1793 1794 1795 1796 1797
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  */
1798 1799 1800 1801 1802 1803

	/*
	 * time inode (stat data) was last changed
	 * (except changes to sd_atime and sd_mtime)
	 */
	__le32 sd_ctime;
1804 1805 1806 1807 1808 1809
	__le32 sd_blocks;
	union {
		__le32 sd_rdev;
		__le32 sd_generation;
	} __attribute__ ((__packed__)) u;
} __attribute__ ((__packed__));
1810 1811

/* this is 44 bytes long */
1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841
#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((sdp)->sd_gid))
#define set_sd_v2_gid(sdp,v)    ((sdp)->sd_gid = cpu_to_le32(v))
#define sd_v2_atime(sdp)        (le32_to_cpu((sdp)->sd_atime))
#define set_sd_v2_atime(sdp,v)  ((sdp)->sd_atime = cpu_to_le32(v))
#define sd_v2_mtime(sdp)        (le32_to_cpu((sdp)->sd_mtime))
#define set_sd_v2_mtime(sdp,v)  ((sdp)->sd_mtime = cpu_to_le32(v))
#define sd_v2_ctime(sdp)        (le32_to_cpu((sdp)->sd_ctime))
#define set_sd_v2_ctime(sdp,v)  ((sdp)->sd_ctime = cpu_to_le32(v))
#define sd_v2_blocks(sdp)       (le32_to_cpu((sdp)->sd_blocks))
#define set_sd_v2_blocks(sdp,v) ((sdp)->sd_blocks = cpu_to_le32(v))
#define sd_v2_rdev(sdp)         (le32_to_cpu((sdp)->u.sd_rdev))
#define set_sd_v2_rdev(sdp,v)   ((sdp)->u.sd_rdev = cpu_to_le32(v))
#define sd_v2_generation(sdp)   (le32_to_cpu((sdp)->u.sd_generation))
#define set_sd_v2_generation(sdp,v) ((sdp)->u.sd_generation = cpu_to_le32(v))
#define sd_v2_attrs(sdp)         (le16_to_cpu((sdp)->sd_attrs))
#define set_sd_v2_attrs(sdp,v)   ((sdp)->sd_attrs = cpu_to_le16(v))

1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858
/***************************************************************************
 *                      DIRECTORY STRUCTURE                                *
 ***************************************************************************/
/*
 * Picture represents the structure of directory items
 * ________________________________________________
 * |  Array of     |   |     |        |       |   |
 * | directory     |N-1| N-2 | ....   |   1st |0th|
 * | entry headers |   |     |        |       |   |
 * |_______________|___|_____|________|_______|___|
 *                  <----   directory entries         ------>
 *
 * First directory item has k_offset component 1. We store "." and ".."
 * in one item, always, we never split "." and ".." into differing
 * items.  This makes, among other things, the code for removing
 * directories simpler.
 */
1859 1860 1861 1862 1863 1864 1865 1866 1867
#define SD_OFFSET  0
#define SD_UNIQUENESS 0
#define DOT_OFFSET 1
#define DOT_DOT_OFFSET 2
#define DIRENTRY_UNIQUENESS 500

#define FIRST_ITEM_OFFSET 1

/*
1868 1869 1870 1871 1872
 * Q: How to get key of object pointed to by entry from entry?
 *
 * A: Each directory entry has its header. This header has deh_dir_id
 *    and deh_objectid fields, those are key of object, entry points to
 */
1873

1874 1875 1876 1877
/*
 * NOT IMPLEMENTED:
 * Directory will someday contain stat data of object
 */
1878 1879 1880

struct reiserfs_de_head {
	__le32 deh_offset;	/* third component of the directory entry key */
1881 1882 1883 1884 1885 1886 1887 1888 1889

	/*
	 * objectid of the parent directory of the object, that is referenced
	 * by directory entry
	 */
	__le32 deh_dir_id;

	/* objectid of the object, that is referenced by directory entry */
	__le32 deh_objectid;
1890
	__le16 deh_location;	/* offset of name in the whole item */
1891 1892 1893 1894 1895 1896

	/*
	 * whether 1) entry contains stat data (for future), and
	 * 2) whether entry is hidden (unlinked)
	 */
	__le16 deh_state;
1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925
} __attribute__ ((__packed__));
#define DEH_SIZE                  sizeof(struct reiserfs_de_head)
#define deh_offset(p_deh)         (le32_to_cpu((p_deh)->deh_offset))
#define deh_dir_id(p_deh)         (le32_to_cpu((p_deh)->deh_dir_id))
#define deh_objectid(p_deh)       (le32_to_cpu((p_deh)->deh_objectid))
#define deh_location(p_deh)       (le16_to_cpu((p_deh)->deh_location))
#define deh_state(p_deh)          (le16_to_cpu((p_deh)->deh_state))

#define put_deh_offset(p_deh,v)   ((p_deh)->deh_offset = cpu_to_le32((v)))
#define put_deh_dir_id(p_deh,v)   ((p_deh)->deh_dir_id = cpu_to_le32((v)))
#define put_deh_objectid(p_deh,v) ((p_deh)->deh_objectid = cpu_to_le32((v)))
#define put_deh_location(p_deh,v) ((p_deh)->deh_location = cpu_to_le16((v)))
#define put_deh_state(p_deh,v)    ((p_deh)->deh_state = cpu_to_le16((v)))

/* empty directory contains two entries "." and ".." and their headers */
#define EMPTY_DIR_SIZE \
(DEH_SIZE * 2 + ROUND_UP (strlen (".")) + ROUND_UP (strlen ("..")))

/* old format directories have this size when empty */
#define EMPTY_DIR_SIZE_V1 (DEH_SIZE * 2 + 3)

#define DEH_Statdata 0		/* not used now */
#define DEH_Visible 2

/* 64 bit systems (and the S/390) need to be aligned explicitly -jdm */
#if BITS_PER_LONG == 64 || defined(__s390__) || defined(__hppa__)
#   define ADDR_UNALIGNED_BITS  (3)
#endif

1926 1927
/*
 * These are only used to manipulate deh_state.
1928
 * Because of this, we'll use the ext2_ bit routines,
1929 1930
 * since they are little endian
 */
1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964
#ifdef ADDR_UNALIGNED_BITS

#   define aligned_address(addr)           ((void *)((long)(addr) & ~((1UL << ADDR_UNALIGNED_BITS) - 1)))
#   define unaligned_offset(addr)          (((int)((long)(addr) & ((1 << ADDR_UNALIGNED_BITS) - 1))) << 3)

#   define set_bit_unaligned(nr, addr)	\
	__test_and_set_bit_le((nr) + unaligned_offset(addr), aligned_address(addr))
#   define clear_bit_unaligned(nr, addr)	\
	__test_and_clear_bit_le((nr) + unaligned_offset(addr), aligned_address(addr))
#   define test_bit_unaligned(nr, addr)	\
	test_bit_le((nr) + unaligned_offset(addr), aligned_address(addr))

#else

#   define set_bit_unaligned(nr, addr)	__test_and_set_bit_le(nr, addr)
#   define clear_bit_unaligned(nr, addr)	__test_and_clear_bit_le(nr, addr)
#   define test_bit_unaligned(nr, addr)	test_bit_le(nr, addr)

#endif

#define mark_de_with_sd(deh)        set_bit_unaligned (DEH_Statdata, &((deh)->deh_state))
#define mark_de_without_sd(deh)     clear_bit_unaligned (DEH_Statdata, &((deh)->deh_state))
#define mark_de_visible(deh)	    set_bit_unaligned (DEH_Visible, &((deh)->deh_state))
#define mark_de_hidden(deh)	    clear_bit_unaligned (DEH_Visible, &((deh)->deh_state))

#define de_with_sd(deh)		    test_bit_unaligned (DEH_Statdata, &((deh)->deh_state))
#define de_visible(deh)	    	    test_bit_unaligned (DEH_Visible, &((deh)->deh_state))
#define de_hidden(deh)	    	    !test_bit_unaligned (DEH_Visible, &((deh)->deh_state))

extern void make_empty_dir_item_v1(char *body, __le32 dirid, __le32 objid,
				   __le32 par_dirid, __le32 par_objid);
extern void make_empty_dir_item(char *body, __le32 dirid, __le32 objid,
				__le32 par_dirid, __le32 par_objid);

1965
/* two entries per block (at least) */
1966 1967
#define REISERFS_MAX_NAME(block_size) 255

1968 1969 1970 1971 1972 1973 1974
/*
 * this structure is used for operations on directory entries. It is
 * not a disk structure.
 *
 * When reiserfs_find_entry or search_by_entry_key find directory
 * entry, they return filled reiserfs_dir_entry structure
 */
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991
struct reiserfs_dir_entry {
	struct buffer_head *de_bh;
	int de_item_num;
	struct item_head *de_ih;
	int de_entry_num;
	struct reiserfs_de_head *de_deh;
	int de_entrylen;
	int de_namelen;
	char *de_name;
	unsigned long *de_gen_number_bit_string;

	__u32 de_dir_id;
	__u32 de_objectid;

	struct cpu_key de_entry_key;
};

1992 1993 1994 1995
/*
 * these defines are useful when a particular member of
 * a reiserfs_dir_entry is needed
 */
1996 1997

/* pointer to file name, stored in entry */
1998 1999
#define B_I_DEH_ENTRY_FILE_NAME(bh, ih, deh) \
				(ih_item_body(bh, ih) + deh_location(deh))
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021

/* length of name */
#define I_DEH_N_ENTRY_FILE_NAME_LENGTH(ih,deh,entry_num) \
(I_DEH_N_ENTRY_LENGTH (ih, deh, entry_num) - (de_with_sd (deh) ? SD_SIZE : 0))

/* hash value occupies bits from 7 up to 30 */
#define GET_HASH_VALUE(offset) ((offset) & 0x7fffff80LL)
/* generation number occupies 7 bits starting from 0 up to 6 */
#define GET_GENERATION_NUMBER(offset) ((offset) & 0x7fLL)
#define MAX_GENERATION_NUMBER  127

#define SET_GENERATION_NUMBER(offset,gen_number) (GET_HASH_VALUE(offset)|(gen_number))

/*
 * Picture represents an internal node of the reiserfs tree
 *  ______________________________________________________
 * |      |  Array of     |  Array of         |  Free     |
 * |block |    keys       |  pointers         | space     |
 * | head |      N        |      N+1          |           |
 * |______|_______________|___________________|___________|
 */

2022 2023 2024 2025 2026 2027 2028
/***************************************************************************
 *                      DISK CHILD                                         *
 ***************************************************************************/
/*
 * Disk child pointer:
 * The pointer from an internal node of the tree to a node that is on disk.
 */
2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060
struct disk_child {
	__le32 dc_block_number;	/* Disk child's block number. */
	__le16 dc_size;		/* Disk child's used space.   */
	__le16 dc_reserved;
};

#define DC_SIZE (sizeof(struct disk_child))
#define dc_block_number(dc_p)	(le32_to_cpu((dc_p)->dc_block_number))
#define dc_size(dc_p)		(le16_to_cpu((dc_p)->dc_size))
#define put_dc_block_number(dc_p, val)   do { (dc_p)->dc_block_number = cpu_to_le32(val); } while(0)
#define put_dc_size(dc_p, val)   do { (dc_p)->dc_size = cpu_to_le16(val); } while(0)

/* Get disk child by buffer header and position in the tree node. */
#define B_N_CHILD(bh, n_pos)  ((struct disk_child *)\
((bh)->b_data + BLKH_SIZE + B_NR_ITEMS(bh) * KEY_SIZE + DC_SIZE * (n_pos)))

/* Get disk child number by buffer header and position in the tree node. */
#define B_N_CHILD_NUM(bh, n_pos) (dc_block_number(B_N_CHILD(bh, n_pos)))
#define PUT_B_N_CHILD_NUM(bh, n_pos, val) \
				(put_dc_block_number(B_N_CHILD(bh, n_pos), val))

 /* maximal value of field child_size in structure disk_child */
 /* child size is the combined size of all items and their headers */
#define MAX_CHILD_SIZE(bh) ((int)( (bh)->b_size - BLKH_SIZE ))

/* amount of used space in buffer (not including block head) */
#define B_CHILD_SIZE(cur) (MAX_CHILD_SIZE(cur)-(B_FREE_SPACE(cur)))

/* max and min number of keys in internal node */
#define MAX_NR_KEY(bh) ( (MAX_CHILD_SIZE(bh)-DC_SIZE)/(KEY_SIZE+DC_SIZE) )
#define MIN_NR_KEY(bh)    (MAX_NR_KEY(bh)/2)

2061 2062 2063
/***************************************************************************
 *                      PATH STRUCTURES AND DEFINES                        *
 ***************************************************************************/
2064

2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076
/*
 * search_by_key fills up the path from the root to the leaf as it descends
 * the tree looking for the key.  It uses reiserfs_bread to try to find
 * buffers in the cache given their block number.  If it does not find
 * them in the cache it reads them from disk.  For each node search_by_key
 * finds using reiserfs_bread it then uses bin_search to look through that
 * node.  bin_search will find the position of the block_number of the next
 * node if it is looking through an internal node.  If it is looking through
 * a leaf node bin_search will find the position of the item which has key
 * either equal to given key, or which is the maximal key less than the
 * given key.
 */
2077 2078

struct path_element {
2079 2080 2081 2082
	/* Pointer to the buffer at the path in the tree. */
	struct buffer_head *pe_buffer;
	/* Position in the tree node which is placed in the buffer above. */
	int pe_position;
2083 2084
};

2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120
/*
 * maximal height of a tree. don't change this without
 * changing JOURNAL_PER_BALANCE_CNT
 */
#define MAX_HEIGHT 5

/* Must be equals MAX_HEIGHT + FIRST_PATH_ELEMENT_OFFSET */
#define EXTENDED_MAX_HEIGHT         7

/* Must be equal to at least 2. */
#define FIRST_PATH_ELEMENT_OFFSET   2

/* Must be equal to FIRST_PATH_ELEMENT_OFFSET - 1 */
#define ILLEGAL_PATH_ELEMENT_OFFSET 1

/* this MUST be MAX_HEIGHT + 1. See about FEB below */
#define MAX_FEB_SIZE 6

/*
 * We need to keep track of who the ancestors of nodes are.  When we
 * perform a search we record which nodes were visited while
 * descending the tree looking for the node we searched for. This list
 * of nodes is called the path.  This information is used while
 * performing balancing.  Note that this path information may become
 * invalid, and this means we must check it when using it to see if it
 * is still valid. You'll need to read search_by_key and the comments
 * in it, especially about decrement_counters_in_path(), to understand
 * this structure.
 *
 * Paths make the code so much harder to work with and debug.... An
 * enormous number of bugs are due to them, and trying to write or modify
 * code that uses them just makes my head hurt.  They are based on an
 * excessive effort to avoid disturbing the precious VFS code.:-( The
 * gods only know how we are going to SMP the code that uses them.
 * znodes are the way!
 */
2121 2122 2123 2124 2125 2126 2127

#define PATH_READA	0x1	/* do read ahead */
#define PATH_READA_BACK 0x2	/* read backwards */

struct treepath {
	int path_length;	/* Length of the array above.   */
	int reada;
2128 2129
	/* Array of the path elements.  */
	struct path_element path_elements[EXTENDED_MAX_HEIGHT];
2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147
	int pos_in_item;
};

#define pos_in_item(path) ((path)->pos_in_item)

#define INITIALIZE_PATH(var) \
struct treepath var = {.path_length = ILLEGAL_PATH_ELEMENT_OFFSET, .reada = 0,}

/* Get path element by path and path position. */
#define PATH_OFFSET_PELEMENT(path, n_offset)  ((path)->path_elements + (n_offset))

/* Get buffer header at the path by path and path position. */
#define PATH_OFFSET_PBUFFER(path, n_offset)   (PATH_OFFSET_PELEMENT(path, n_offset)->pe_buffer)

/* Get position in the element at the path by path and path position. */
#define PATH_OFFSET_POSITION(path, n_offset) (PATH_OFFSET_PELEMENT(path, n_offset)->pe_position)

#define PATH_PLAST_BUFFER(path) (PATH_OFFSET_PBUFFER((path), (path)->path_length))
2148 2149 2150 2151 2152 2153

/*
 * you know, to the person who didn't write this the macro name does not
 * at first suggest what it does.  Maybe POSITION_FROM_PATH_END? Or
 * maybe we should just focus on dumping paths... -Hans
 */
2154 2155
#define PATH_LAST_POSITION(path) (PATH_OFFSET_POSITION((path), (path)->path_length))

2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172
/*
 * in do_balance leaf has h == 0 in contrast with path structure,
 * where root has level == 0. That is why we need these defines
 */

/* tb->S[h] */
#define PATH_H_PBUFFER(path, h) \
			PATH_OFFSET_PBUFFER(path, path->path_length - (h))

/* tb->F[h] or tb->S[0]->b_parent */
#define PATH_H_PPARENT(path, h) PATH_H_PBUFFER(path, (h) + 1)

#define PATH_H_POSITION(path, h) \
			PATH_OFFSET_POSITION(path, path->path_length - (h))

/* tb->S[h]->b_item_order */
#define PATH_H_B_ITEM_ORDER(path, h) PATH_H_POSITION(path, h + 1)
2173 2174 2175

#define PATH_H_PATH_OFFSET(path, n_h) ((path)->path_length - (n_h))

2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227
static inline void *reiserfs_node_data(const struct buffer_head *bh)
{
	return bh->b_data + sizeof(struct block_head);
}

/* get key from internal node */
static inline struct reiserfs_key *internal_key(struct buffer_head *bh,
						int item_num)
{
	struct reiserfs_key *key = reiserfs_node_data(bh);

	return &key[item_num];
}

/* get the item header from leaf node */
static inline struct item_head *item_head(const struct buffer_head *bh,
					  int item_num)
{
	struct item_head *ih = reiserfs_node_data(bh);

	return &ih[item_num];
}

/* get the key from leaf node */
static inline struct reiserfs_key *leaf_key(const struct buffer_head *bh,
					    int item_num)
{
	return &item_head(bh, item_num)->ih_key;
}

static inline void *ih_item_body(const struct buffer_head *bh,
				 const struct item_head *ih)
{
	return bh->b_data + ih_location(ih);
}

/* get item body from leaf node */
static inline void *item_body(const struct buffer_head *bh, int item_num)
{
	return ih_item_body(bh, item_head(bh, item_num));
}

static inline struct item_head *tp_item_head(const struct treepath *path)
{
	return item_head(PATH_PLAST_BUFFER(path), PATH_LAST_POSITION(path));
}

static inline void *tp_item_body(const struct treepath *path)
{
	return item_body(PATH_PLAST_BUFFER(path), PATH_LAST_POSITION(path));
}

2228 2229 2230 2231 2232
#define get_last_bh(path) PATH_PLAST_BUFFER(path)
#define get_item_pos(path) PATH_LAST_POSITION(path)
#define item_moved(ih,path) comp_items(ih, path)
#define path_changed(ih,path) comp_items (ih, path)

2233 2234 2235 2236 2237
/* array of the entry headers */
 /* get item body */
#define B_I_DEH(bh, ih) ((struct reiserfs_de_head *)(ih_item_body(bh, ih)))

/*
2238 2239 2240 2241 2242 2243 2244
 * length of the directory entry in directory item. This define
 * calculates length of i-th directory entry using directory entry
 * locations from dir entry head. When it calculates length of 0-th
 * directory entry, it uses length of whole item in place of entry
 * location of the non-existent following entry in the calculation.
 * See picture above.
 */
2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
static inline int entry_length(const struct buffer_head *bh,
			       const struct item_head *ih, int pos_in_item)
{
	struct reiserfs_de_head *deh;

	deh = B_I_DEH(bh, ih) + pos_in_item;
	if (pos_in_item)
		return deh_location(deh - 1) - deh_location(deh);

	return ih_item_len(ih) - deh_location(deh);
}

2257 2258 2259
/***************************************************************************
 *                       MISC                                              *
 ***************************************************************************/
2260 2261 2262 2263 2264

/* Size of pointer to the unformatted node. */
#define UNFM_P_SIZE (sizeof(unp_t))
#define UNFM_P_SHIFT 2

2265
/* in in-core inode key is stored on le form */
2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288
#define INODE_PKEY(inode) ((struct reiserfs_key *)(REISERFS_I(inode)->i_key))

#define MAX_UL_INT 0xffffffff
#define MAX_INT    0x7ffffff
#define MAX_US_INT 0xffff

// reiserfs version 2 has max offset 60 bits. Version 1 - 32 bit offset
static inline loff_t max_reiserfs_offset(struct inode *inode)
{
	if (get_inode_item_key_version(inode) == KEY_FORMAT_3_5)
		return (loff_t) U32_MAX;

	return (loff_t) ((~(__u64) 0) >> 4);
}

#define MAX_KEY_OBJECTID	MAX_UL_INT

#define MAX_B_NUM  MAX_UL_INT
#define MAX_FC_NUM MAX_US_INT

/* the purpose is to detect overflow of an unsigned short */
#define REISERFS_LINK_MAX (MAX_US_INT - 1000)

2289 2290 2291 2292 2293 2294
/*
 * The following defines are used in reiserfs_insert_item
 * and reiserfs_append_item
 */
#define REISERFS_KERNEL_MEM		0	/* kernel memory mode */
#define REISERFS_USER_MEM		1	/* user memory mode */
2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305

#define fs_generation(s) (REISERFS_SB(s)->s_generation_counter)
#define get_generation(s) atomic_read (&fs_generation(s))
#define FILESYSTEM_CHANGED_TB(tb)  (get_generation((tb)->tb_sb) != (tb)->fs_gen)
#define __fs_changed(gen,s) (gen != get_generation (s))
#define fs_changed(gen,s)		\
({					\
	reiserfs_cond_resched(s);	\
	__fs_changed(gen, s);		\
})

2306 2307 2308
/***************************************************************************
 *                  FIXATE NODES                                           *
 ***************************************************************************/
2309 2310 2311 2312

#define VI_TYPE_LEFT_MERGEABLE 1
#define VI_TYPE_RIGHT_MERGEABLE 2

2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327
/*
 * To make any changes in the tree we always first find node, that
 * contains item to be changed/deleted or place to insert a new
 * item. We call this node S. To do balancing we need to decide what
 * we will shift to left/right neighbor, or to a new node, where new
 * item will be etc. To make this analysis simpler we build virtual
 * node. Virtual node is an array of items, that will replace items of
 * node S. (For instance if we are going to delete an item, virtual
 * node does not contain it). Virtual node keeps information about
 * item sizes and types, mergeability of first and last items, sizes
 * of all entries in directory item. We use this array of items when
 * calculating what we can shift to neighbors and how many nodes we
 * have to have if we do not any shiftings, if we shift to left/right
 * neighbor or to both.
 */
2328
struct virtual_item {
2329 2330 2331 2332 2333 2334
	int vi_index;		/* index in the array of item operations */
	unsigned short vi_type;	/* left/right mergeability */

	/* length of item that it will have after balancing */
	unsigned short vi_item_len;

2335
	struct item_head *vi_ih;
2336 2337 2338
	const char *vi_item;	/* body of item (old or new) */
	const void *vi_new_data;	/* 0 always but paste mode */
	void *vi_uarea;		/* item specific area */
2339 2340 2341
};

struct virtual_node {
2342 2343 2344
	/* this is a pointer to the free space in the buffer */
	char *vn_free_ptr;

2345
	unsigned short vn_nr_item;	/* number of items in virtual node */
2346 2347 2348 2349 2350 2351 2352 2353 2354 2355

	/*
	 * size of node , that node would have if it has
	 * unlimited size and no balancing is performed
	 */
	short vn_size;

	/* mode of balancing (paste, insert, delete, cut) */
	short vn_mode;

2356 2357
	short vn_affected_item_num;
	short vn_pos_in_item;