reiserfs_fs.h 92.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5
/*
 * Copyright 1996, 1997, 1998 Hans Reiser, see reiserfs/README for licensing and copyright details
 */

				/* this file has an amazingly stupid
6 7 8 9
				   name, yura please fix it to be
				   reiserfs.h, and merge all the rest
				   of our .h files that are in this
				   directory into it.  */
L
Linus Torvalds 已提交
10 11 12 13 14

#ifndef _LINUX_REISER_FS_H
#define _LINUX_REISER_FS_H

#include <linux/types.h>
15 16
#include <linux/magic.h>

L
Linus Torvalds 已提交
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37
#ifdef __KERNEL__
#include <linux/slab.h>
#include <linux/interrupt.h>
#include <linux/sched.h>
#include <linux/workqueue.h>
#include <asm/unaligned.h>
#include <linux/bitops.h>
#include <linux/proc_fs.h>
#include <linux/smp_lock.h>
#include <linux/buffer_head.h>
#include <linux/reiserfs_fs_i.h>
#include <linux/reiserfs_fs_sb.h>
#endif

/*
 *  include/linux/reiser_fs.h
 *
 *  Reiser File System constants and structures
 *
 */

38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
/* ioctl's command */
#define REISERFS_IOC_UNPACK		_IOW(0xCD,1,long)
/* define following flags to be the same as in ext2, so that chattr(1),
   lsattr(1) will work with us. */
#define REISERFS_IOC_GETFLAGS		FS_IOC_GETFLAGS
#define REISERFS_IOC_SETFLAGS		FS_IOC_SETFLAGS
#define REISERFS_IOC_GETVERSION		FS_IOC_GETVERSION
#define REISERFS_IOC_SETVERSION		FS_IOC_SETVERSION

#ifdef __KERNEL__
/* 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

F
Frederic Weisbecker 已提交
55 56 57 58 59 60 61
/*
 * 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);
62 63
int reiserfs_write_lock_once(struct super_block *s);
void reiserfs_write_unlock_once(struct super_block *s, int lock_depth);
64

65 66 67 68 69 70
#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

71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
/*
 * Several mutexes depend on the write lock.
 * However sometimes we want to relax the write lock while we hold
 * these mutexes, according to the release/reacquire on schedule()
 * properties of the Bkl that were used.
 * Reiserfs performances and locking were based on this scheme.
 * Now that the write lock is a mutex and not the bkl anymore, doing so
 * may result in a deadlock:
 *
 * A acquire write_lock
 * A acquire j_commit_mutex
 * A release write_lock and wait for something
 * B acquire write_lock
 * B can't acquire j_commit_mutex and sleep
 * A can't acquire write lock anymore
 * deadlock
 *
 * What we do here is avoiding such deadlock by playing the same game
 * than the Bkl: if we can't acquire a mutex that depends on the write lock,
 * we release the write lock, wait a bit and then retry.
 *
 * The mutexes concerned by this hack are:
 * - The commit mutex of a journal list
 * - The flush mutex
 * - The journal lock
 * - The inode mutex
 */
static inline void reiserfs_mutex_lock_safe(struct mutex *m,
			       struct super_block *s)
{
101
	reiserfs_lock_check_recursive(s);
102 103 104 105 106
	reiserfs_write_unlock(s);
	mutex_lock(m);
	reiserfs_write_lock(s);
}

107 108 109 110
static inline void
reiserfs_mutex_lock_nested_safe(struct mutex *m, unsigned int subclass,
			       struct super_block *s)
{
111
	reiserfs_lock_check_recursive(s);
112 113 114 115 116
	reiserfs_write_unlock(s);
	mutex_lock_nested(m, subclass);
	reiserfs_write_lock(s);
}

117 118 119
static inline void
reiserfs_down_read_safe(struct rw_semaphore *sem, struct super_block *s)
{
120
	reiserfs_lock_check_recursive(s);
121 122 123 124 125
	reiserfs_write_unlock(s);
	down_read(sem);
	reiserfs_write_lock(s);
}

126 127 128 129 130 131 132 133 134 135 136 137 138
/*
 * When we schedule, we usually want to also release the write lock,
 * according to the previous bkl based locking scheme of reiserfs.
 */
static inline void reiserfs_cond_resched(struct super_block *s)
{
	if (need_resched()) {
		reiserfs_write_unlock(s);
		schedule();
		reiserfs_write_lock(s);
	}
}

139 140
struct fid;

L
Linus Torvalds 已提交
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
/* in reading the #defines, it may help to understand that they employ
   the following abbreviations:

   B = Buffer
   I = Item header
   H = Height within the tree (should be changed to LEV)
   N = Number of the item in the node
   STAT = stat data
   DEH = Directory Entry Header
   EC = Entry Count
   E = Entry number
   UL = Unsigned Long
   BLKH = BLocK Header
   UNFM = UNForMatted node
   DC = Disk Child
   P = Path

   These #defines are named by concatenating these abbreviations,
   where first comes the arguments, and last comes the return value,
   of the macro.

*/

#define USE_INODE_GENERATION_COUNTER

#define REISERFS_PREALLOCATE
#define DISPLACE_NEW_PACKING_LOCALITIES
#define PREALLOCATION_SIZE 9

/* n must be power of 2 */
#define _ROUND_UP(x,n) (((x)+(n)-1u) & ~((n)-1u))

// to be ok for alpha and others we have to align structures to 8 byte
// boundary.
// FIXME: do not change 4 by anything else: there is code which relies on that
#define ROUND_UP(x) _ROUND_UP(x,8LL)

/* debug levels.  Right now, CONFIG_REISERFS_CHECK means print all debug
** messages.
*/
181
#define REISERFS_DEBUG_CODE 5	/* extra messages to help find/debug errors */
L
Linus Torvalds 已提交
182

J
Jeff Mahoney 已提交
183 184 185 186
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)
L
Linus Torvalds 已提交
187 188 189
/* assertions handling */

/** always check a condition and panic if it's false. */
J
Jeff Mahoney 已提交
190 191 192 193 194 195 196 197
#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)
L
Linus Torvalds 已提交
198

A
Al Viro 已提交
199 200
#define RASSERT(cond, format, args...) __RASSERT(cond, #cond, format, ##args)

L
Linus Torvalds 已提交
201
#if defined( CONFIG_REISERFS_CHECK )
A
Al Viro 已提交
202
#define RFALSE(cond, format, args...) __RASSERT(!(cond), "!(" #cond ")", format, ##args)
L
Linus Torvalds 已提交
203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
#else
#define RFALSE( cond, format, args... ) do {;} while( 0 )
#endif

#define CONSTF __attribute_const__
/*
 * Disk Data Structures
 */

/***************************************************************************/
/*                             SUPER BLOCK                                 */
/***************************************************************************/

/*
 * Structure of super block on disk, a version of which in RAM is often accessed as REISERFS_SB(s)->s_rs
 * the version in RAM is part of a larger structure containing fields never written to disk.
 */
220 221
#define UNSET_HASH 0		// read_super will guess about, what hash names
		     // in directories were sorted with
L
Linus Torvalds 已提交
222 223 224 225 226 227
#define TEA_HASH  1
#define YURA_HASH 2
#define R5_HASH   3
#define DEFAULT_HASH R5_HASH

struct journal_params {
228 229 230 231 232 233 234 235 236 237 238 239 240
	__le32 jp_journal_1st_block;	/* where does journal start from on its
					 * device */
	__le32 jp_journal_dev;	/* journal device st_rdev */
	__le32 jp_journal_size;	/* size of the journal */
	__le32 jp_journal_trans_max;	/* max number of blocks in a transaction. */
	__le32 jp_journal_magic;	/* random value made on fs creation (this
					 * was sb_journal_block_count) */
	__le32 jp_journal_max_batch;	/* max number of blocks to batch into a
					 * trans */
	__le32 jp_journal_max_commit_age;	/* in seconds, how old can an async
						 * commit be */
	__le32 jp_journal_max_trans_age;	/* in seconds, how old can a transaction
						 * be */
L
Linus Torvalds 已提交
241 242 243
};

/* this is the super from 3.5.X, where X >= 10 */
244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269
struct reiserfs_super_block_v1 {
	__le32 s_block_count;	/* blocks count         */
	__le32 s_free_blocks;	/* free blocks count    */
	__le32 s_root_block;	/* root block number    */
	struct journal_params s_journal;
	__le16 s_blocksize;	/* block size */
	__le16 s_oid_maxsize;	/* max size of object id array, see
				 * get_objectid() commentary  */
	__le16 s_oid_cursize;	/* current size of object id array */
	__le16 s_umount_state;	/* this is set to 1 when filesystem was
				 * umounted, to 2 - when not */
	char s_magic[10];	/* reiserfs magic string indicates that
				 * file system is reiserfs:
				 * "ReIsErFs" or "ReIsEr2Fs" or "ReIsEr3Fs" */
	__le16 s_fs_state;	/* it is set to used by fsck to mark which
				 * phase of rebuilding is done */
	__le32 s_hash_function_code;	/* indicate, what hash function is being use
					 * to sort names in a directory*/
	__le16 s_tree_height;	/* height of disk tree */
	__le16 s_bmap_nr;	/* amount of bitmap blocks needed to address
				 * each block of file system */
	__le16 s_version;	/* this field is only reliable on filesystem
				 * with non-standard journal */
	__le16 s_reserved_for_journal;	/* size in blocks of journal area on main
					 * device, we need to keep after
					 * making fs with non-standard journal */
L
Linus Torvalds 已提交
270 271 272 273 274
} __attribute__ ((__packed__));

#define SB_SIZE_V1 (sizeof(struct reiserfs_super_block_v1))

/* this is the on disk super block */
275 276 277 278 279 280
struct reiserfs_super_block {
	struct reiserfs_super_block_v1 s_v1;
	__le32 s_inode_generation;
	__le32 s_flags;		/* Right now used only by inode-attributes, if enabled */
	unsigned char s_uuid[16];	/* filesystem unique identifier */
	unsigned char s_label[16];	/* filesystem volume label */
281 282 283 284 285
	__le16 s_mnt_count;		/* Count of mounts since last fsck */
	__le16 s_max_mnt_count;		/* Maximum mounts before check */
	__le32 s_lastcheck;		/* Timestamp of last fsck */
	__le32 s_check_interval;	/* Interval between checks */
	char s_unused[76];	/* zero filled by mkreiserfs and
286 287 288 289
				 * reiserfs_convert_objectid_map_v1()
				 * so any additions must be updated
				 * there as well. */
} __attribute__ ((__packed__));
L
Linus Torvalds 已提交
290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324

#define SB_SIZE (sizeof(struct reiserfs_super_block))

#define REISERFS_VERSION_1 0
#define REISERFS_VERSION_2 2

// on-disk super block fields converted to cpu form
#define SB_DISK_SUPER_BLOCK(s) (REISERFS_SB(s)->s_rs)
#define SB_V1_DISK_SUPER_BLOCK(s) (&(SB_DISK_SUPER_BLOCK(s)->s_v1))
#define SB_BLOCKSIZE(s) \
        le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_blocksize))
#define SB_BLOCK_COUNT(s) \
        le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_block_count))
#define SB_FREE_BLOCKS(s) \
        le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_free_blocks))
#define SB_REISERFS_MAGIC(s) \
        (SB_V1_DISK_SUPER_BLOCK(s)->s_magic)
#define SB_ROOT_BLOCK(s) \
        le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_root_block))
#define SB_TREE_HEIGHT(s) \
        le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_tree_height))
#define SB_REISERFS_STATE(s) \
        le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_umount_state))
#define SB_VERSION(s) le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_version))
#define SB_BMAP_NR(s) le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_bmap_nr))

#define PUT_SB_BLOCK_COUNT(s, val) \
   do { SB_V1_DISK_SUPER_BLOCK(s)->s_block_count = cpu_to_le32(val); } while (0)
#define PUT_SB_FREE_BLOCKS(s, val) \
   do { SB_V1_DISK_SUPER_BLOCK(s)->s_free_blocks = cpu_to_le32(val); } while (0)
#define PUT_SB_ROOT_BLOCK(s, val) \
   do { SB_V1_DISK_SUPER_BLOCK(s)->s_root_block = cpu_to_le32(val); } while (0)
#define PUT_SB_TREE_HEIGHT(s, val) \
   do { SB_V1_DISK_SUPER_BLOCK(s)->s_tree_height = cpu_to_le16(val); } while (0)
#define PUT_SB_REISERFS_STATE(s, val) \
325
   do { SB_V1_DISK_SUPER_BLOCK(s)->s_umount_state = cpu_to_le16(val); } while (0)
L
Linus Torvalds 已提交
326 327 328 329 330 331 332 333 334 335 336 337 338
#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) \
339
         le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_reserved_for_journal))
L
Linus Torvalds 已提交
340 341 342 343 344

#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)) ? \
345
         SB_ONDISK_JOURNAL_SIZE(s) + 1 : SB_ONDISK_RESERVED_FOR_JOURNAL(s)))
L
Linus Torvalds 已提交
346

347 348 349
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);
L
Linus Torvalds 已提交
350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372

/* ReiserFS leaves the first 64k unused, so that partition labels have
   enough space.  If someone wants to write a fancy bootloader that
   needs more than 64k, let us know, and this will be increased in size.
   This number must be larger than than the largest block size on any
   platform, or code will break.  -Hans */
#define REISERFS_DISK_OFFSET_IN_BYTES (64 * 1024)
#define REISERFS_FIRST_BLOCK unused_define
#define REISERFS_JOURNAL_OFFSET_IN_BYTES REISERFS_DISK_OFFSET_IN_BYTES

/* the spot for the super in versions 3.5 - 3.5.10 (inclusive) */
#define REISERFS_OLD_DISK_OFFSET_IN_BYTES (8 * 1024)

// reiserfs internal error code (used by search_by_key adn 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;
373
typedef __le32 unp_t;
L
Linus Torvalds 已提交
374 375

struct unfm_nodeinfo {
376 377
	unp_t unfm_nodenum;
	unsigned short unfm_freespace;
L
Linus Torvalds 已提交
378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398
};

/* there are two formats of keys: 3.5 and 3.6
 */
#define KEY_FORMAT_3_5 0
#define KEY_FORMAT_3_6 1

/* there are two stat datas */
#define STAT_DATA_V1 0
#define STAT_DATA_V2 1

static inline struct reiserfs_inode_info *REISERFS_I(const struct inode *inode)
{
	return container_of(inode, struct reiserfs_inode_info, vfs_inode);
}

static inline struct reiserfs_sb_info *REISERFS_SB(const struct super_block *sb)
{
	return sb->s_fs_info;
}

399 400
/* Don't trust REISERFS_SB(sb)->s_bmap_nr, it's a u16
 * which overflows on large file systems. */
401
static inline __u32 reiserfs_bmap_count(struct super_block *sb)
402 403 404 405 406 407 408 409 410
{
	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);
}

L
Linus Torvalds 已提交
411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
/** this says about version of key of all items (but stat data) the
    object consists of */
#define get_inode_item_key_version( inode )                                    \
    ((REISERFS_I(inode)->i_flags & i_item_key_version_mask) ? KEY_FORMAT_3_6 : KEY_FORMAT_3_5)

#define set_inode_item_key_version( inode, version )                           \
         ({ if((version)==KEY_FORMAT_3_6)                                      \
                REISERFS_I(inode)->i_flags |= i_item_key_version_mask;      \
            else                                                               \
                REISERFS_I(inode)->i_flags &= ~i_item_key_version_mask; })

#define get_inode_sd_version(inode)                                            \
    ((REISERFS_I(inode)->i_flags & i_stat_data_version_mask) ? STAT_DATA_V2 : STAT_DATA_V1)

#define set_inode_sd_version(inode, version)                                   \
         ({ if((version)==STAT_DATA_V2)                                        \
                REISERFS_I(inode)->i_flags |= i_stat_data_version_mask;     \
            else                                                               \
                REISERFS_I(inode)->i_flags &= ~i_stat_data_version_mask; })

/* This is an aggressive tail suppression policy, I am hoping it
   improves our benchmarks. The principle behind it is that percentage
   space saving is what matters, not absolute space saving.  This is
   non-intuitive, but it helps to understand it if you consider that the
   cost to access 4 blocks is not much more than the cost to access 1
   block, if you have to do a seek and rotate.  A tail risks a
   non-linear disk access that is significant as a percentage of total
   time cost for a 4 block file and saves an amount of space that is
   less significant as a percentage of space, or so goes the hypothesis.
   -Hans */
#define STORE_TAIL_IN_UNFM_S1(n_file_size,n_tail_size,n_block_size) \
(\
  (!(n_tail_size)) || \
  (((n_tail_size) > MAX_DIRECT_ITEM_LEN(n_block_size)) || \
   ( (n_file_size) >= (n_block_size) * 4 ) || \
   ( ( (n_file_size) >= (n_block_size) * 3 ) && \
     ( (n_tail_size) >=   (MAX_DIRECT_ITEM_LEN(n_block_size))/4) ) || \
   ( ( (n_file_size) >= (n_block_size) * 2 ) && \
     ( (n_tail_size) >=   (MAX_DIRECT_ITEM_LEN(n_block_size))/2) ) || \
   ( ( (n_file_size) >= (n_block_size) ) && \
     ( (n_tail_size) >=   (MAX_DIRECT_ITEM_LEN(n_block_size) * 3)/4) ) ) \
)

/* Another strategy for tails, this one means only create a tail if all the
   file would fit into one DIRECT item.
   Primary intention for this one is to increase performance by decreasing
   seeking.
458
*/
L
Linus Torvalds 已提交
459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476
#define STORE_TAIL_IN_UNFM_S2(n_file_size,n_tail_size,n_block_size) \
(\
  (!(n_tail_size)) || \
  (((n_file_size) > MAX_DIRECT_ITEM_LEN(n_block_size)) ) \
)

/*
 * values for s_umount_state field
 */
#define REISERFS_VALID_FS    1
#define REISERFS_ERROR_FS    2

//
// there are 5 item types currently
//
#define TYPE_STAT_DATA 0
#define TYPE_INDIRECT 1
#define TYPE_DIRECT 2
477 478 479
#define TYPE_DIRENTRY 3
#define TYPE_MAXTYPE 3
#define TYPE_ANY 15		// FIXME: comment is required
L
Linus Torvalds 已提交
480 481 482 483 484 485 486 487 488

/***************************************************************************/
/*                       KEY & ITEM HEAD                                   */
/***************************************************************************/

//
// directories use this key as well as old files
//
struct offset_v1 {
489 490
	__le32 k_offset;
	__le32 k_uniqueness;
L
Linus Torvalds 已提交
491 492 493
} __attribute__ ((__packed__));

struct offset_v2 {
494
	__le64 v;
L
Linus Torvalds 已提交
495 496
} __attribute__ ((__packed__));

497
static inline __u16 offset_v2_k_type(const struct offset_v2 *v2)
L
Linus Torvalds 已提交
498
{
499
	__u8 type = le64_to_cpu(v2->v) >> 60;
500
	return (type <= TYPE_MAXTYPE) ? type : TYPE_ANY;
L
Linus Torvalds 已提交
501
}
502 503

static inline void set_offset_v2_k_type(struct offset_v2 *v2, int type)
L
Linus Torvalds 已提交
504
{
505 506
	v2->v =
	    (v2->v & cpu_to_le64(~0ULL >> 4)) | cpu_to_le64((__u64) type << 60);
L
Linus Torvalds 已提交
507
}
508 509

static inline loff_t offset_v2_k_offset(const struct offset_v2 *v2)
L
Linus Torvalds 已提交
510
{
511
	return le64_to_cpu(v2->v) & (~0ULL >> 4);
L
Linus Torvalds 已提交
512 513
}

514 515 516 517
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);
L
Linus Torvalds 已提交
518 519 520 521 522
}

/* Key of an item determines its location in the S+tree, and
   is composed of 4 components */
struct reiserfs_key {
523 524 525 526 527 528 529
	__le32 k_dir_id;	/* packing locality: by default parent
				   directory object id */
	__le32 k_objectid;	/* object identifier */
	union {
		struct offset_v1 k_offset_v1;
		struct offset_v2 k_offset_v2;
	} __attribute__ ((__packed__)) u;
L
Linus Torvalds 已提交
530 531
} __attribute__ ((__packed__));

532
struct in_core_key {
533 534 535 536 537
	__u32 k_dir_id;		/* packing locality: by default parent
				   directory object id */
	__u32 k_objectid;	/* object identifier */
	__u64 k_offset;
	__u8 k_type;
538
};
L
Linus Torvalds 已提交
539 540

struct cpu_key {
541 542 543 544
	struct in_core_key on_disk_key;
	int version;
	int key_length;		/* 3 in all cases but direct2indirect and
				   indirect2direct conversion */
L
Linus Torvalds 已提交
545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589
};

/* Our function for comparing keys can compare keys of different
   lengths.  It takes as a parameter the length of the keys it is to
   compare.  These defines are used in determining what is to be passed
   to it as that parameter. */
#define REISERFS_FULL_KEY_LEN     4
#define REISERFS_SHORT_KEY_LEN    2

/* The result of the key compare */
#define FIRST_GREATER 1
#define SECOND_GREATER -1
#define KEYS_IDENTICAL 0
#define KEY_FOUND 1
#define KEY_NOT_FOUND 0

#define KEY_SIZE (sizeof(struct reiserfs_key))
#define SHORT_KEY_SIZE (sizeof (__u32) + sizeof (__u32))

/* return values for search_by_key and clones */
#define ITEM_FOUND 1
#define ITEM_NOT_FOUND 0
#define ENTRY_FOUND 1
#define ENTRY_NOT_FOUND 0
#define DIRECTORY_NOT_FOUND -1
#define REGULAR_FILE_FOUND -2
#define DIRECTORY_FOUND -3
#define BYTE_FOUND 1
#define BYTE_NOT_FOUND 0
#define FILE_NOT_FOUND -1

#define POSITION_FOUND 1
#define POSITION_NOT_FOUND 0

// return values for reiserfs_find_entry and search_by_entry_key
#define NAME_FOUND 1
#define NAME_NOT_FOUND 0
#define GOTO_PREVIOUS_ITEM 2
#define NAME_FOUND_INVISIBLE 3

/*  Everything in the filesystem is stored as a set of items.  The
    item head contains the key of the item, its free space (for
    indirect items) and specifies the location of the item itself
    within the block.  */

590
struct item_head {
L
Linus Torvalds 已提交
591 592 593 594 595 596 597 598 599 600
	/* Everything in the tree is found by searching for it based on
	 * its key.*/
	struct reiserfs_key ih_key;
	union {
		/* The free space in the last unformatted node of an
		   indirect item if this is an indirect item.  This
		   equals 0xFFFF iff this is a direct item or stat data
		   item. Note that the key, not this field, is used to
		   determine the item type, and thus which field this
		   union contains. */
601
		__le16 ih_free_space_reserved;
L
Linus Torvalds 已提交
602 603
		/* Iff this is a directory item, this field equals the
		   number of directory entries in the directory item. */
604
		__le16 ih_entry_count;
L
Linus Torvalds 已提交
605
	} __attribute__ ((__packed__)) u;
606 607 608 609 610 611 612
	__le16 ih_item_len;	/* total size of the item body */
	__le16 ih_item_location;	/* an offset to the item body
					 * within the block */
	__le16 ih_version;	/* 0 for all old items, 2 for new
				   ones. Highest bit is set by fsck
				   temporary, cleaned after all
				   done */
L
Linus Torvalds 已提交
613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639
} __attribute__ ((__packed__));
/* size of item header     */
#define IH_SIZE (sizeof(struct item_head))

#define ih_free_space(ih)            le16_to_cpu((ih)->u.ih_free_space_reserved)
#define ih_version(ih)               le16_to_cpu((ih)->ih_version)
#define ih_entry_count(ih)           le16_to_cpu((ih)->u.ih_entry_count)
#define ih_location(ih)              le16_to_cpu((ih)->ih_item_location)
#define ih_item_len(ih)              le16_to_cpu((ih)->ih_item_len)

#define put_ih_free_space(ih, val)   do { (ih)->u.ih_free_space_reserved = cpu_to_le16(val); } while(0)
#define put_ih_version(ih, val)      do { (ih)->ih_version = cpu_to_le16(val); } while (0)
#define put_ih_entry_count(ih, val)  do { (ih)->u.ih_entry_count = cpu_to_le16(val); } while (0)
#define put_ih_location(ih, val)     do { (ih)->ih_item_location = cpu_to_le16(val); } while (0)
#define put_ih_item_len(ih, val)     do { (ih)->ih_item_len = cpu_to_le16(val); } while (0)

#define unreachable_item(ih) (ih_version(ih) & (1 << 15))

#define get_ih_free_space(ih) (ih_version (ih) == KEY_FORMAT_3_6 ? 0 : ih_free_space (ih))
#define set_ih_free_space(ih,val) put_ih_free_space((ih), ((ih_version(ih) == KEY_FORMAT_3_6) ? 0 : (val)))

/* these operate on indirect items, where you've got an array of ints
** at a possibly unaligned location.  These are a noop on ia32
** 
** p is the array of __u32, i is the index into the array, v is the value
** to store there.
*/
640 641
#define get_block_num(p, i) get_unaligned_le32((p) + (i))
#define put_block_num(p, i, v) put_unaligned_le32((v), (p) + (i))
L
Linus Torvalds 已提交
642 643 644 645 646 647 648 649

//
// in old version uniqueness field shows key type
//
#define V1_SD_UNIQUENESS 0
#define V1_INDIRECT_UNIQUENESS 0xfffffffe
#define V1_DIRECT_UNIQUENESS 0xffffffff
#define V1_DIRENTRY_UNIQUENESS 500
650
#define V1_ANY_UNIQUENESS 555	// FIXME: comment is required
L
Linus Torvalds 已提交
651 652 653 654

//
// here are conversion routines
//
655 656
static inline int uniqueness2type(__u32 uniqueness) CONSTF;
static inline int uniqueness2type(__u32 uniqueness)
L
Linus Torvalds 已提交
657
{
658 659 660 661 662 663 664 665 666
	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;
L
Linus Torvalds 已提交
667
	case V1_ANY_UNIQUENESS:
668
	default:
669 670
		return TYPE_ANY;
	}
L
Linus Torvalds 已提交
671 672
}

673 674
static inline __u32 type2uniqueness(int type) CONSTF;
static inline __u32 type2uniqueness(int type)
L
Linus Torvalds 已提交
675
{
676 677 678 679 680 681 682 683 684
	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;
L
Linus Torvalds 已提交
685
	case TYPE_ANY:
686
	default:
687 688
		return V1_ANY_UNIQUENESS;
	}
L
Linus Torvalds 已提交
689 690 691 692 693 694 695
}

//
// 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
//
696 697
static inline loff_t le_key_k_offset(int version,
				     const struct reiserfs_key *key)
L
Linus Torvalds 已提交
698
{
699 700 701
	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));
L
Linus Torvalds 已提交
702 703
}

704
static inline loff_t le_ih_k_offset(const struct item_head *ih)
L
Linus Torvalds 已提交
705
{
706
	return le_key_k_offset(ih_version(ih), &(ih->ih_key));
L
Linus Torvalds 已提交
707 708
}

709
static inline loff_t le_key_k_type(int version, const struct reiserfs_key *key)
L
Linus Torvalds 已提交
710
{
711 712 713
	return (version == KEY_FORMAT_3_5) ?
	    uniqueness2type(le32_to_cpu(key->u.k_offset_v1.k_uniqueness)) :
	    offset_v2_k_type(&(key->u.k_offset_v2));
L
Linus Torvalds 已提交
714 715
}

716
static inline loff_t le_ih_k_type(const struct item_head *ih)
L
Linus Torvalds 已提交
717
{
718
	return le_key_k_type(ih_version(ih), &(ih->ih_key));
L
Linus Torvalds 已提交
719 720
}

721 722
static inline void set_le_key_k_offset(int version, struct reiserfs_key *key,
				       loff_t offset)
L
Linus Torvalds 已提交
723
{
724 725
	(version == KEY_FORMAT_3_5) ? (void)(key->u.k_offset_v1.k_offset = cpu_to_le32(offset)) :	/* jdm check */
	    (void)(set_offset_v2_k_offset(&(key->u.k_offset_v2), offset));
L
Linus Torvalds 已提交
726 727
}

728
static inline void set_le_ih_k_offset(struct item_head *ih, loff_t offset)
L
Linus Torvalds 已提交
729
{
730
	set_le_key_k_offset(ih_version(ih), &(ih->ih_key), offset);
L
Linus Torvalds 已提交
731 732
}

733 734
static inline void set_le_key_k_type(int version, struct reiserfs_key *key,
				     int type)
L
Linus Torvalds 已提交
735
{
736 737 738 739
	(version == KEY_FORMAT_3_5) ?
	    (void)(key->u.k_offset_v1.k_uniqueness =
		   cpu_to_le32(type2uniqueness(type)))
	    : (void)(set_offset_v2_k_type(&(key->u.k_offset_v2), type));
L
Linus Torvalds 已提交
740
}
741

742
static inline void set_le_ih_k_type(struct item_head *ih, int type)
L
Linus Torvalds 已提交
743
{
744
	set_le_key_k_type(ih_version(ih), &(ih->ih_key), type);
L
Linus Torvalds 已提交
745 746
}

747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765
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;
}
L
Linus Torvalds 已提交
766 767 768 769

//
// item header has version.
//
770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788
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);
}
L
Linus Torvalds 已提交
789 790 791 792

//
// key is pointer to cpu key, result is cpu
//
793
static inline loff_t cpu_key_k_offset(const struct cpu_key *key)
L
Linus Torvalds 已提交
794
{
795
	return key->on_disk_key.k_offset;
L
Linus Torvalds 已提交
796 797
}

798
static inline loff_t cpu_key_k_type(const struct cpu_key *key)
L
Linus Torvalds 已提交
799
{
800
	return key->on_disk_key.k_type;
L
Linus Torvalds 已提交
801 802
}

803
static inline void set_cpu_key_k_offset(struct cpu_key *key, loff_t offset)
L
Linus Torvalds 已提交
804
{
805
	key->on_disk_key.k_offset = offset;
L
Linus Torvalds 已提交
806 807
}

808
static inline void set_cpu_key_k_type(struct cpu_key *key, int type)
L
Linus Torvalds 已提交
809
{
810
	key->on_disk_key.k_type = type;
L
Linus Torvalds 已提交
811 812
}

813
static inline void cpu_key_k_offset_dec(struct cpu_key *key)
L
Linus Torvalds 已提交
814
{
815
	key->on_disk_key.k_offset--;
L
Linus Torvalds 已提交
816 817 818 819 820 821 822 823 824 825 826 827 828
}

#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)))

J
Jeff Mahoney 已提交
829 830 831
#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))
L
Linus Torvalds 已提交
832

833
/* maximal length of item */
L
Linus Torvalds 已提交
834 835 836 837 838 839
#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
840

L
Linus Torvalds 已提交
841 842 843 844 845 846 847 848 849 850 851 852 853
extern struct reiserfs_key root_key;

/* 
 * Picture represents a leaf of the S+tree
 *  ______________________________________________________
 * |      |  Array of     |                   |           |
 * |Block |  Object-Item  |      F r e e      |  Objects- |
 * | head |  Headers      |     S p a c e     |   Items   |
 * |______|_______________|___________________|___________|
 */

/* Header of a disk block.  More precisely, header of a formatted leaf
   or internal node, and not the header of an unformatted node. */
854 855 856 857 858 859 860
struct block_head {
	__le16 blk_level;	/* Level of a block in the tree. */
	__le16 blk_nr_item;	/* Number of keys/items in a block. */
	__le16 blk_free_space;	/* Block free space in bytes. */
	__le16 blk_reserved;
	/* dump this in v4/planA */
	struct reiserfs_key blk_right_delim_key;	/* kept only for compatibility */
L
Linus Torvalds 已提交
861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878
};

#define BLKH_SIZE                     (sizeof(struct block_head))
#define blkh_level(p_blkh)            (le16_to_cpu((p_blkh)->blk_level))
#define blkh_nr_item(p_blkh)          (le16_to_cpu((p_blkh)->blk_nr_item))
#define blkh_free_space(p_blkh)       (le16_to_cpu((p_blkh)->blk_free_space))
#define blkh_reserved(p_blkh)         (le16_to_cpu((p_blkh)->blk_reserved))
#define set_blkh_level(p_blkh,val)    ((p_blkh)->blk_level = cpu_to_le16(val))
#define set_blkh_nr_item(p_blkh,val)  ((p_blkh)->blk_nr_item = cpu_to_le16(val))
#define set_blkh_free_space(p_blkh,val) ((p_blkh)->blk_free_space = cpu_to_le16(val))
#define set_blkh_reserved(p_blkh,val) ((p_blkh)->blk_reserved = cpu_to_le16(val))
#define blkh_right_delim_key(p_blkh)  ((p_blkh)->blk_right_delim_key)
#define set_blkh_right_delim_key(p_blkh,val)  ((p_blkh)->blk_right_delim_key = val)

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

879 880 881 882
#define FREE_LEVEL 0		/* when node gets removed from the tree its
				   blk_level is set to FREE_LEVEL. It is then
				   used to see whether the node is still in the
				   tree */
L
Linus Torvalds 已提交
883

884
#define DISK_LEAF_NODE_LEVEL  1	/* Leaf node level. */
L
Linus Torvalds 已提交
885 886

/* Given the buffer head of a formatted node, resolve to the block head of that node. */
J
Jeff Mahoney 已提交
887
#define B_BLK_HEAD(bh)			((struct block_head *)((bh)->b_data))
L
Linus Torvalds 已提交
888
/* Number of items that are in buffer. */
J
Jeff Mahoney 已提交
889 890 891
#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)))
L
Linus Torvalds 已提交
892

J
Jeff Mahoney 已提交
893 894 895
#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)
L
Linus Torvalds 已提交
896 897

/* Get right delimiting key. -- little endian */
J
Jeff Mahoney 已提交
898
#define B_PRIGHT_DELIM_KEY(bh)		(&(blk_right_delim_key(B_BLK_HEAD(bh))))
L
Linus Torvalds 已提交
899 900

/* Does the buffer contain a disk leaf. */
J
Jeff Mahoney 已提交
901
#define B_IS_ITEMS_LEVEL(bh)		(B_LEVEL(bh) == DISK_LEAF_NODE_LEVEL)
L
Linus Torvalds 已提交
902 903

/* Does the buffer contain a disk internal node */
J
Jeff Mahoney 已提交
904 905
#define B_IS_KEYS_LEVEL(bh)      (B_LEVEL(bh) > DISK_LEAF_NODE_LEVEL \
					    && B_LEVEL(bh) <= MAX_HEIGHT)
L
Linus Torvalds 已提交
906 907 908 909 910 911 912 913 914

/***************************************************************************/
/*                             STAT DATA                                   */
/***************************************************************************/

//
// old stat data is 32 bytes long. We are going to distinguish new one by
// different size
//
915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936
struct stat_data_v1 {
	__le16 sd_mode;		/* file type, permissions */
	__le16 sd_nlink;	/* number of hard links */
	__le16 sd_uid;		/* owner */
	__le16 sd_gid;		/* group */
	__le32 sd_size;		/* file size */
	__le32 sd_atime;	/* time of last access */
	__le32 sd_mtime;	/* time file was last modified  */
	__le32 sd_ctime;	/* time inode (stat data) was last changed (except changes to sd_atime and sd_mtime) */
	union {
		__le32 sd_rdev;
		__le32 sd_blocks;	/* number of blocks file uses */
	} __attribute__ ((__packed__)) u;
	__le32 sd_first_direct_byte;	/* first byte of file which is stored
					   in a direct item: except that if it
					   equals 1 it is a symlink and if it
					   equals ~(__u32)0 there is no
					   direct item.  The existence of this
					   field really grates on me. Let's
					   replace it with a macro based on
					   sd_size and our tail suppression
					   policy.  Someday.  -Hans */
L
Linus Torvalds 已提交
937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969
} __attribute__ ((__packed__));

#define SD_V1_SIZE              (sizeof(struct stat_data_v1))
#define stat_data_v1(ih)        (ih_version (ih) == KEY_FORMAT_3_5)
#define sd_v1_mode(sdp)         (le16_to_cpu((sdp)->sd_mode))
#define set_sd_v1_mode(sdp,v)   ((sdp)->sd_mode = cpu_to_le16(v))
#define sd_v1_nlink(sdp)        (le16_to_cpu((sdp)->sd_nlink))
#define set_sd_v1_nlink(sdp,v)  ((sdp)->sd_nlink = cpu_to_le16(v))
#define sd_v1_uid(sdp)          (le16_to_cpu((sdp)->sd_uid))
#define set_sd_v1_uid(sdp,v)    ((sdp)->sd_uid = cpu_to_le16(v))
#define sd_v1_gid(sdp)          (le16_to_cpu((sdp)->sd_gid))
#define set_sd_v1_gid(sdp,v)    ((sdp)->sd_gid = cpu_to_le16(v))
#define sd_v1_size(sdp)         (le32_to_cpu((sdp)->sd_size))
#define set_sd_v1_size(sdp,v)   ((sdp)->sd_size = cpu_to_le32(v))
#define sd_v1_atime(sdp)        (le32_to_cpu((sdp)->sd_atime))
#define set_sd_v1_atime(sdp,v)  ((sdp)->sd_atime = cpu_to_le32(v))
#define sd_v1_mtime(sdp)        (le32_to_cpu((sdp)->sd_mtime))
#define set_sd_v1_mtime(sdp,v)  ((sdp)->sd_mtime = cpu_to_le32(v))
#define sd_v1_ctime(sdp)        (le32_to_cpu((sdp)->sd_ctime))
#define set_sd_v1_ctime(sdp,v)  ((sdp)->sd_ctime = cpu_to_le32(v))
#define sd_v1_rdev(sdp)         (le32_to_cpu((sdp)->u.sd_rdev))
#define set_sd_v1_rdev(sdp,v)   ((sdp)->u.sd_rdev = cpu_to_le32(v))
#define sd_v1_blocks(sdp)       (le32_to_cpu((sdp)->u.sd_blocks))
#define set_sd_v1_blocks(sdp,v) ((sdp)->u.sd_blocks = cpu_to_le32(v))
#define sd_v1_first_direct_byte(sdp) \
                                (le32_to_cpu((sdp)->sd_first_direct_byte))
#define set_sd_v1_first_direct_byte(sdp,v) \
                                ((sdp)->sd_first_direct_byte = cpu_to_le32(v))

/* inode flags stored in sd_attrs (nee sd_reserved) */

/* we want common flags to have the same values as in ext2,
   so chattr(1) will work without problems */
970 971 972 973 974 975 976 977 978
#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
L
Linus Torvalds 已提交
979 980 981 982 983 984 985 986 987 988 989 990 991

/* persistent flags that file inherits from the parent directory */
#define REISERFS_INHERIT_MASK ( REISERFS_IMMUTABLE_FL |	\
				REISERFS_SYNC_FL |	\
				REISERFS_NOATIME_FL |	\
				REISERFS_NODUMP_FL |	\
				REISERFS_SECRM_FL |	\
				REISERFS_COMPR_FL |	\
				REISERFS_NOTAIL_FL )

/* Stat Data on disk (reiserfs version of UFS disk inode minus the
   address blocks) */
struct stat_data {
992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014
	__le16 sd_mode;		/* file type, permissions */
	__le16 sd_attrs;	/* persistent inode flags */
	__le32 sd_nlink;	/* number of hard links */
	__le64 sd_size;		/* file size */
	__le32 sd_uid;		/* owner */
	__le32 sd_gid;		/* group */
	__le32 sd_atime;	/* time of last access */
	__le32 sd_mtime;	/* time file was last modified  */
	__le32 sd_ctime;	/* time inode (stat data) was last changed (except changes to sd_atime and sd_mtime) */
	__le32 sd_blocks;
	union {
		__le32 sd_rdev;
		__le32 sd_generation;
		//__le32 sd_first_direct_byte;
		/* first byte of file which is stored in a
		   direct item: except that if it equals 1
		   it is a symlink and if it equals
		   ~(__u32)0 there is no direct item.  The
		   existence of this field really grates
		   on me. Let's replace it with a macro
		   based on sd_size and our tail
		   suppression policy? */
	} __attribute__ ((__packed__)) u;
L
Linus Torvalds 已提交
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 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
} __attribute__ ((__packed__));
//
// this is 44 bytes long
//
#define SD_SIZE (sizeof(struct stat_data))
#define SD_V2_SIZE              SD_SIZE
#define stat_data_v2(ih)        (ih_version (ih) == KEY_FORMAT_3_6)
#define sd_v2_mode(sdp)         (le16_to_cpu((sdp)->sd_mode))
#define set_sd_v2_mode(sdp,v)   ((sdp)->sd_mode = cpu_to_le16(v))
/* sd_reserved */
/* set_sd_reserved */
#define sd_v2_nlink(sdp)        (le32_to_cpu((sdp)->sd_nlink))
#define set_sd_v2_nlink(sdp,v)  ((sdp)->sd_nlink = cpu_to_le32(v))
#define sd_v2_size(sdp)         (le64_to_cpu((sdp)->sd_size))
#define set_sd_v2_size(sdp,v)   ((sdp)->sd_size = cpu_to_le64(v))
#define sd_v2_uid(sdp)          (le32_to_cpu((sdp)->sd_uid))
#define set_sd_v2_uid(sdp,v)    ((sdp)->sd_uid = cpu_to_le32(v))
#define sd_v2_gid(sdp)          (le32_to_cpu((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))

/***************************************************************************/
/*                      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. */
#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

/*
   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 */

/* NOT IMPLEMENTED:   
   Directory will someday contain stat data of object */

1083 1084 1085 1086 1087 1088 1089 1090
struct reiserfs_de_head {
	__le32 deh_offset;	/* third component of the directory entry key */
	__le32 deh_dir_id;	/* objectid of the parent directory of the object, that is referenced
				   by directory entry */
	__le32 deh_objectid;	/* objectid of the object, that is referenced by directory entry */
	__le16 deh_location;	/* offset of name in the whole item */
	__le16 deh_state;	/* whether 1) entry contains stat data (for future), and 2) whether
				   entry is hidden (unlinked) */
L
Linus Torvalds 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
} __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)

1112
#define DEH_Statdata 0		/* not used now */
L
Linus Torvalds 已提交
1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148
#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

/* These are only used to manipulate deh_state.
 * Because of this, we'll use the ext2_ bit routines,
 * since they are little endian */
#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)     ext2_set_bit((nr) + unaligned_offset(addr), aligned_address(addr))
#   define clear_bit_unaligned(nr, addr)   ext2_clear_bit((nr) + unaligned_offset(addr), aligned_address(addr))
#   define test_bit_unaligned(nr, addr)    ext2_test_bit((nr) + unaligned_offset(addr), aligned_address(addr))

#else

#   define set_bit_unaligned(nr, addr)     ext2_set_bit(nr, addr)
#   define clear_bit_unaligned(nr, addr)   ext2_clear_bit(nr, addr)
#   define test_bit_unaligned(nr, addr)    ext2_test_bit(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))

1149 1150 1151 1152
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);
L
Linus Torvalds 已提交
1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168

/* array of the entry headers */
 /* get item body */
#define B_I_PITEM(bh,ih) ( (bh)->b_data + ih_location(ih) )
#define B_I_DEH(bh,ih) ((struct reiserfs_de_head *)(B_I_PITEM(bh,ih)))

/* 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.*/
/*
#define I_DEH_N_ENTRY_LENGTH(ih,deh,i) \
((i) ? (deh_location((deh)-1) - deh_location((deh))) : (ih_item_len((ih)) - deh_location((deh))))
*/
1169 1170
static inline int entry_length(const struct buffer_head *bh,
			       const struct item_head *ih, int pos_in_item)
L
Linus Torvalds 已提交
1171
{
1172
	struct reiserfs_de_head *deh;
L
Linus Torvalds 已提交
1173

1174 1175 1176
	deh = B_I_DEH(bh, ih) + pos_in_item;
	if (pos_in_item)
		return deh_location(deh - 1) - deh_location(deh);
L
Linus Torvalds 已提交
1177

1178
	return ih_item_len(ih) - deh_location(deh);
L
Linus Torvalds 已提交
1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
}

/* number of entries in the directory item, depends on ENTRY_COUNT being at the start of directory dynamic data. */
#define I_ENTRY_COUNT(ih) (ih_entry_count((ih)))

/* name by bh, ih and entry_num */
#define B_I_E_NAME(bh,ih,entry_num) ((char *)(bh->b_data + ih_location(ih) + deh_location(B_I_DEH(bh,ih)+(entry_num))))

// two entries per block (at least)
#define REISERFS_MAX_NAME(block_size) 255

/* 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 */
1194 1195 1196 1197 1198 1199 1200 1201 1202
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;
1203
	unsigned long *de_gen_number_bit_string;
1204 1205 1206 1207 1208

	__u32 de_dir_id;
	__u32 de_objectid;

	struct cpu_key de_entry_key;
L
Linus Torvalds 已提交
1209
};
1210

L
Linus Torvalds 已提交
1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242
/* these defines are useful when a particular member of a reiserfs_dir_entry is needed */

/* pointer to file name, stored in entry */
#define B_I_DEH_ENTRY_FILE_NAME(bh,ih,deh) (B_I_PITEM (bh, ih) + deh_location(deh))

/* 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          |           |
 * |______|_______________|___________________|___________|
 */

/***************************************************************************/
/*                      DISK CHILD                                         */
/***************************************************************************/
/* Disk child pointer: The pointer from an internal node of the tree
   to a node that is on disk. */
struct disk_child {
1243 1244 1245
	__le32 dc_block_number;	/* Disk child's block number. */
	__le16 dc_size;		/* Disk child's used space.   */
	__le16 dc_reserved;
L
Linus Torvalds 已提交
1246 1247 1248 1249 1250 1251 1252 1253 1254
};

#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. */
J
Jeff Mahoney 已提交
1255 1256
#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)))
L
Linus Torvalds 已提交
1257 1258

/* Get disk child number by buffer header and position in the tree node. */
J
Jeff Mahoney 已提交
1259 1260 1261
#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))
L
Linus Torvalds 已提交
1262

1263
 /* maximal value of field child_size in structure disk_child */
L
Linus Torvalds 已提交
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
 /* 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)

/***************************************************************************/
/*                      PATH STRUCTURES AND DEFINES                        */
/***************************************************************************/

/* 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. */

1286 1287 1288 1289
struct path_element {
	struct buffer_head *pe_buffer;	/* Pointer to the buffer at the path in the tree. */
	int pe_position;	/* Position in the tree node which is placed in the */
	/* buffer above.                                  */
L
Linus Torvalds 已提交
1290 1291
};

1292 1293 1294
#define MAX_HEIGHT 5		/* maximal height of a tree. don't change this without changing JOURNAL_PER_BALANCE_CNT */
#define EXTENDED_MAX_HEIGHT         7	/* Must be equals MAX_HEIGHT + FIRST_PATH_ELEMENT_OFFSET */
#define FIRST_PATH_ELEMENT_OFFSET   2	/* Must be equal to at least 2. */
L
Linus Torvalds 已提交
1295

1296 1297
#define ILLEGAL_PATH_ELEMENT_OFFSET 1	/* Must be equal to FIRST_PATH_ELEMENT_OFFSET - 1 */
#define MAX_FEB_SIZE 6		/* this MUST be MAX_HEIGHT + 1. See about FEB below */
L
Linus Torvalds 已提交
1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315

/* 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! */

1316 1317
#define PATH_READA	0x1	/* do read ahead */
#define PATH_READA_BACK 0x2	/* read backwards */
L
Linus Torvalds 已提交
1318

1319
struct treepath {
1320 1321 1322 1323
	int path_length;	/* Length of the array above.   */
	int reada;
	struct path_element path_elements[EXTENDED_MAX_HEIGHT];	/* Array of the path elements.  */
	int pos_in_item;
L
Linus Torvalds 已提交
1324 1325 1326 1327 1328
};

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

#define INITIALIZE_PATH(var) \
1329
struct treepath var = {.path_length = ILLEGAL_PATH_ELEMENT_OFFSET, .reada = 0,}
L
Linus Torvalds 已提交
1330 1331

/* Get path element by path and path position. */
J
Jeff Mahoney 已提交
1332
#define PATH_OFFSET_PELEMENT(path, n_offset)  ((path)->path_elements + (n_offset))
L
Linus Torvalds 已提交
1333 1334

/* Get buffer header at the path by path and path position. */
J
Jeff Mahoney 已提交
1335
#define PATH_OFFSET_PBUFFER(path, n_offset)   (PATH_OFFSET_PELEMENT(path, n_offset)->pe_buffer)
L
Linus Torvalds 已提交
1336 1337

/* Get position in the element at the path by path and path position. */
J
Jeff Mahoney 已提交
1338
#define PATH_OFFSET_POSITION(path, n_offset) (PATH_OFFSET_PELEMENT(path, n_offset)->pe_position)
L
Linus Torvalds 已提交
1339

J
Jeff Mahoney 已提交
1340
#define PATH_PLAST_BUFFER(path) (PATH_OFFSET_PBUFFER((path), (path)->path_length))
L
Linus Torvalds 已提交
1341
				/* you know, to the person who didn't
1342 1343 1344 1345 1346
				   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 */
J
Jeff Mahoney 已提交
1347
#define PATH_LAST_POSITION(path) (PATH_OFFSET_POSITION((path), (path)->path_length))
L
Linus Torvalds 已提交
1348

J
Jeff Mahoney 已提交
1349
#define PATH_PITEM_HEAD(path)    B_N_PITEM_HEAD(PATH_PLAST_BUFFER(path), PATH_LAST_POSITION(path))
L
Linus Torvalds 已提交
1350 1351 1352

/* in do_balance leaf has h == 0 in contrast with path structure,
   where root has level == 0. That is why we need these defines */
J
Jeff Mahoney 已提交
1353
#define PATH_H_PBUFFER(path, h) PATH_OFFSET_PBUFFER (path, path->path_length - (h))	/* tb->S[h] */
1354 1355 1356
#define PATH_H_PPARENT(path, h) PATH_H_PBUFFER (path, (h) + 1)	/* tb->F[h] or tb->S[0]->b_parent */
#define PATH_H_POSITION(path, h) PATH_OFFSET_POSITION (path, path->path_length - (h))
#define PATH_H_B_ITEM_ORDER(path, h) PATH_H_POSITION(path, h + 1)	/* tb->S[h]->b_item_order */
L
Linus Torvalds 已提交
1357

J
Jeff Mahoney 已提交
1358
#define PATH_H_PATH_OFFSET(path, n_h) ((path)->path_length - (n_h))
L
Linus Torvalds 已提交
1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384

#define get_last_bh(path) PATH_PLAST_BUFFER(path)
#define get_ih(path) PATH_PITEM_HEAD(path)
#define get_item_pos(path) PATH_LAST_POSITION(path)
#define get_item(path) ((void *)B_N_PITEM(PATH_PLAST_BUFFER(path), PATH_LAST_POSITION (path)))
#define item_moved(ih,path) comp_items(ih, path)
#define path_changed(ih,path) comp_items (ih, path)

/***************************************************************************/
/*                       MISC                                              */
/***************************************************************************/

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

// in in-core inode key is stored on le form
#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
#define U32_MAX (~(__u32)0)

1385
static inline loff_t max_reiserfs_offset(struct inode *inode)
L
Linus Torvalds 已提交
1386
{
1387 1388
	if (get_inode_item_key_version(inode) == KEY_FORMAT_3_5)
		return (loff_t) U32_MAX;
L
Linus Torvalds 已提交
1389

1390
	return (loff_t) ((~(__u64) 0) >> 4);
L
Linus Torvalds 已提交
1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402
}

/*#define MAX_KEY_UNIQUENESS	MAX_UL_INT*/
#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)

/* The following defines are used in reiserfs_insert_item and reiserfs_append_item  */
1403 1404
#define REISERFS_KERNEL_MEM		0	/* reiserfs kernel memory mode  */
#define REISERFS_USER_MEM		1	/* reiserfs user memory mode            */
L
Linus Torvalds 已提交
1405 1406 1407 1408 1409

#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))
1410 1411
#define fs_changed(gen,s)		\
({					\
1412
	reiserfs_cond_resched(s);	\
1413 1414
	__fs_changed(gen, s);		\
})
L
Linus Torvalds 已提交
1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435

/***************************************************************************/
/*                  FIXATE NODES                                           */
/***************************************************************************/

#define VI_TYPE_LEFT_MERGEABLE 1
#define VI_TYPE_RIGHT_MERGEABLE 2

/* 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. */
1436 1437 1438 1439 1440 1441 1442 1443
struct virtual_item {
	int vi_index;		// index in the array of item operations
	unsigned short vi_type;	// left/right mergeability
	unsigned short vi_item_len;	/* length of item that it will have after balancing */
	struct item_head *vi_ih;
	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
L
Linus Torvalds 已提交
1444 1445
};

1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
struct virtual_node {
	char *vn_free_ptr;	/* this is a pointer to the free space in the buffer */
	unsigned short vn_nr_item;	/* number of items in virtual node */
	short vn_size;		/* size of node , that node would have if it has unlimited size and no balancing is performed */
	short vn_mode;		/* mode of balancing (paste, insert, delete, cut) */
	short vn_affected_item_num;
	short vn_pos_in_item;
	struct item_head *vn_ins_ih;	/* item header of inserted item, 0 for other modes */
	const void *vn_data;
	struct virtual_item *vn_vi;	/* array of items (including a new one, excluding item to be deleted) */
L
Linus Torvalds 已提交
1456 1457 1458 1459
};

/* used by directory items when creating virtual nodes */
struct direntry_uarea {
1460 1461 1462 1463
	int flags;
	__u16 entry_count;
	__u16 entry_sizes[1];
} __attribute__ ((__packed__));
L
Linus Torvalds 已提交
1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485

/***************************************************************************/
/*                  TREE BALANCE                                           */
/***************************************************************************/

/* This temporary structure is used in tree balance algorithms, and
   constructed as we go to the extent that its various parts are
   needed.  It contains arrays of nodes that can potentially be
   involved in the balancing of node S, and parameters that define how
   each of the nodes must be balanced.  Note that in these algorithms
   for balancing the worst case is to need to balance the current node
   S and the left and right neighbors and all of their parents plus
   create a new node.  We implement S1 balancing for the leaf nodes
   and S0 balancing for the internal nodes (S1 and S0 are defined in
   our papers.)*/

#define MAX_FREE_BLOCK 7	/* size of the array of buffers to free at end of do_balance */

/* maximum number of FEB blocknrs on a single level */
#define MAX_AMOUNT_NEEDED 2

/* someday somebody will prefix every field in this struct with tb_ */
1486 1487 1488 1489 1490
struct tree_balance {
	int tb_mode;
	int need_balance_dirty;
	struct super_block *tb_sb;
	struct reiserfs_transaction_handle *transaction_handle;
1491
	struct treepath *tb_path;
1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539
	struct buffer_head *L[MAX_HEIGHT];	/* array of left neighbors of nodes in the path */
	struct buffer_head *R[MAX_HEIGHT];	/* array of right neighbors of nodes in the path */
	struct buffer_head *FL[MAX_HEIGHT];	/* array of fathers of the left  neighbors      */
	struct buffer_head *FR[MAX_HEIGHT];	/* array of fathers of the right neighbors      */
	struct buffer_head *CFL[MAX_HEIGHT];	/* array of common parents of center node and its left neighbor  */
	struct buffer_head *CFR[MAX_HEIGHT];	/* array of common parents of center node and its right neighbor */

	struct buffer_head *FEB[MAX_FEB_SIZE];	/* array of empty buffers. Number of buffers in array equals
						   cur_blknum. */
	struct buffer_head *used[MAX_FEB_SIZE];
	struct buffer_head *thrown[MAX_FEB_SIZE];
	int lnum[MAX_HEIGHT];	/* array of number of items which must be
				   shifted to the left in order to balance the
				   current node; for leaves includes item that
				   will be partially shifted; for internal
				   nodes, it is the number of child pointers
				   rather than items. It includes the new item
				   being created. The code sometimes subtracts
				   one to get the number of wholly shifted
				   items for other purposes. */
	int rnum[MAX_HEIGHT];	/* substitute right for left in comment above */
	int lkey[MAX_HEIGHT];	/* array indexed by height h mapping the key delimiting L[h] and
				   S[h] to its item number within the node CFL[h] */
	int rkey[MAX_HEIGHT];	/* substitute r for l in comment above */
	int insert_size[MAX_HEIGHT];	/* the number of bytes by we are trying to add or remove from
					   S[h]. A negative value means removing.  */
	int blknum[MAX_HEIGHT];	/* number of nodes that will replace node S[h] after
				   balancing on the level h of the tree.  If 0 then S is
				   being deleted, if 1 then S is remaining and no new nodes
				   are being created, if 2 or 3 then 1 or 2 new nodes is
				   being created */

	/* fields that are used only for balancing leaves of the tree */
	int cur_blknum;		/* number of empty blocks having been already allocated                 */
	int s0num;		/* number of items that fall into left most  node when S[0] splits     */
	int s1num;		/* number of items that fall into first  new node when S[0] splits     */
	int s2num;		/* number of items that fall into second new node when S[0] splits     */
	int lbytes;		/* number of bytes which can flow to the left neighbor from the        left    */
	/* most liquid item that cannot be shifted from S[0] entirely         */
	/* if -1 then nothing will be partially shifted */
	int rbytes;		/* number of bytes which will flow to the right neighbor from the right        */
	/* most liquid item that cannot be shifted from S[0] entirely         */
	/* if -1 then nothing will be partially shifted                           */
	int s1bytes;		/* number of bytes which flow to the first  new node when S[0] splits   */
	/* note: if S[0] splits into 3 nodes, then items do not need to be cut  */
	int s2bytes;
	struct buffer_head *buf_to_free[MAX_FREE_BLOCK];	/* buffers which are to be freed after do_balance finishes by unfix_nodes */
	char *vn_buf;		/* kmalloced memory. Used to create
L
Linus Torvalds 已提交
1540 1541
				   virtual node and keep map of
				   dirtied bitmap blocks */
1542 1543
	int vn_buf_size;	/* size of the vn_buf */
	struct virtual_node *tb_vn;	/* VN starts after bitmap of bitmap blocks */
L
Linus Torvalds 已提交
1544

1545 1546
	int fs_gen;		/* saved value of `reiserfs_generation' counter
				   see FILESYSTEM_CHANGED() macro in reiserfs_fs.h */
L
Linus Torvalds 已提交
1547
#ifdef DISPLACE_NEW_PACKING_LOCALITIES
1548 1549
	struct in_core_key key;	/* key pointer, to pass to block allocator or
				   another low-level subsystem */
L
Linus Torvalds 已提交
1550
#endif
1551
};
L
Linus Torvalds 已提交
1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585

/* These are modes of balancing */

/* When inserting an item. */
#define M_INSERT	'i'
/* When inserting into (directories only) or appending onto an already
   existant item. */
#define M_PASTE		'p'
/* When deleting an item. */
#define M_DELETE	'd'
/* When truncating an item or removing an entry from a (directory) item. */
#define M_CUT 		'c'

/* used when balancing on leaf level skipped (in reiserfsck) */
#define M_INTERNAL	'n'

/* When further balancing is not needed, then do_balance does not need
   to be called. */
#define M_SKIP_BALANCING 		's'
#define M_CONVERT	'v'

/* modes of leaf_move_items */
#define LEAF_FROM_S_TO_L 0
#define LEAF_FROM_S_TO_R 1
#define LEAF_FROM_R_TO_L 2
#define LEAF_FROM_L_TO_R 3
#define LEAF_FROM_S_TO_SNEW 4

#define FIRST_TO_LAST 0
#define LAST_TO_FIRST 1

/* used in do_balance for passing parent of node information that has
   been gotten from tb struct */
struct buffer_info {
1586 1587 1588 1589
	struct tree_balance *tb;
	struct buffer_head *bi_bh;
	struct buffer_head *bi_parent;
	int bi_position;
L
Linus Torvalds 已提交
1590 1591
};

J
Jeff Mahoney 已提交
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601
static inline struct super_block *sb_from_tb(struct tree_balance *tb)
{
	return tb ? tb->tb_sb : NULL;
}

static inline struct super_block *sb_from_bi(struct buffer_info *bi)
{
	return bi ? sb_from_tb(bi->tb) : NULL;
}

L
Linus Torvalds 已提交
1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618
/* there are 4 types of items: stat data, directory item, indirect, direct.
+-------------------+------------+--------------+------------+
|	            |  k_offset  | k_uniqueness | mergeable? |
+-------------------+------------+--------------+------------+
|     stat data     |	0        |      0       |   no       |
+-------------------+------------+--------------+------------+
| 1st directory item| DOT_OFFSET |DIRENTRY_UNIQUENESS|   no       | 
| non 1st directory | hash value |              |   yes      |
|     item          |            |              |            |
+-------------------+------------+--------------+------------+
| indirect item     | offset + 1 |TYPE_INDIRECT |   if this is not the first indirect item of the object
+-------------------+------------+--------------+------------+
| direct item       | offset + 1 |TYPE_DIRECT   | if not this is not the first direct item of the object
+-------------------+------------+--------------+------------+
*/

struct item_operations {
1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633
	int (*bytes_number) (struct item_head * ih, int block_size);
	void (*decrement_key) (struct cpu_key *);
	int (*is_left_mergeable) (struct reiserfs_key * ih,
				  unsigned long bsize);
	void (*print_item) (struct item_head *, char *item);
	void (*check_item) (struct item_head *, char *item);

	int (*create_vi) (struct virtual_node * vn, struct virtual_item * vi,
			  int is_affected, int insert_size);
	int (*check_left) (struct virtual_item * vi, int free,
			   int start_skip, int end_skip);
	int (*check_right) (struct virtual_item * vi, int free);
	int (*part_size) (struct virtual_item * vi, int from, int to);
	int (*unit_num) (struct virtual_item * vi);
	void (*print_vi) (struct virtual_item * vi);
L
Linus Torvalds 已提交
1634 1635
};

1636
extern struct item_operations *item_ops[TYPE_ANY + 1];
L
Linus Torvalds 已提交
1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651

#define op_bytes_number(ih,bsize)                    item_ops[le_ih_k_type (ih)]->bytes_number (ih, bsize)
#define op_is_left_mergeable(key,bsize)              item_ops[le_key_k_type (le_key_version (key), key)]->is_left_mergeable (key, bsize)
#define op_print_item(ih,item)                       item_ops[le_ih_k_type (ih)]->print_item (ih, item)
#define op_check_item(ih,item)                       item_ops[le_ih_k_type (ih)]->check_item (ih, item)
#define op_create_vi(vn,vi,is_affected,insert_size)  item_ops[le_ih_k_type ((vi)->vi_ih)]->create_vi (vn,vi,is_affected,insert_size)
#define op_check_left(vi,free,start_skip,end_skip) item_ops[(vi)->vi_index]->check_left (vi, free, start_skip, end_skip)
#define op_check_right(vi,free)                      item_ops[(vi)->vi_index]->check_right (vi, free)
#define op_part_size(vi,from,to)                     item_ops[(vi)->vi_index]->part_size (vi, from, to)
#define op_unit_num(vi)				     item_ops[(vi)->vi_index]->unit_num (vi)
#define op_print_vi(vi)                              item_ops[(vi)->vi_index]->print_vi (vi)

#define COMP_SHORT_KEYS comp_short_keys

/* number of blocks pointed to by the indirect item */
J
Jeff Mahoney 已提交
1652
#define I_UNFM_NUM(ih)	(ih_item_len(ih) / UNFM_P_SIZE)
L
Linus Torvalds 已提交
1653 1654 1655 1656 1657 1658

/* the used space within the unformatted node corresponding to pos within the item pointed to by ih */
#define I_POS_UNFM_SIZE(ih,pos,size) (((pos) == I_UNFM_NUM(ih) - 1 ) ? (size) - ih_free_space(ih) : (size))

/* number of bytes contained by the direct item or the unformatted nodes the indirect item points to */

1659
/* get the item header */
L
Linus Torvalds 已提交
1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689
#define B_N_PITEM_HEAD(bh,item_num) ( (struct item_head * )((bh)->b_data + BLKH_SIZE) + (item_num) )

/* get key */
#define B_N_PDELIM_KEY(bh,item_num) ( (struct reiserfs_key * )((bh)->b_data + BLKH_SIZE) + (item_num) )

/* get the key */
#define B_N_PKEY(bh,item_num) ( &(B_N_PITEM_HEAD(bh,item_num)->ih_key) )

/* get item body */
#define B_N_PITEM(bh,item_num) ( (bh)->b_data + ih_location(B_N_PITEM_HEAD((bh),(item_num))))

/* get the stat data by the buffer header and the item order */
#define B_N_STAT_DATA(bh,nr) \
( (struct stat_data *)((bh)->b_data + ih_location(B_N_PITEM_HEAD((bh),(nr))) ) )

    /* following defines use reiserfs buffer header and item header */

/* get stat-data */
#define B_I_STAT_DATA(bh, ih) ( (struct stat_data * )((bh)->b_data + ih_location(ih)) )

// this is 3976 for size==4096
#define MAX_DIRECT_ITEM_LEN(size) ((size) - BLKH_SIZE - 2*IH_SIZE - SD_SIZE - UNFM_P_SIZE)

/* indirect items consist of entries which contain blocknrs, pos
   indicates which entry, and B_I_POS_UNFM_POINTER resolves to the
   blocknr contained by the entry pos points to */
#define B_I_POS_UNFM_POINTER(bh,ih,pos) le32_to_cpu(*(((unp_t *)B_I_PITEM(bh,ih)) + (pos)))
#define PUT_B_I_POS_UNFM_POINTER(bh,ih,pos, val) do {*(((unp_t *)B_I_PITEM(bh,ih)) + (pos)) = cpu_to_le32(val); } while (0)

struct reiserfs_iget_args {
1690 1691 1692
	__u32 objectid;
	__u32 dirid;
};
L
Linus Torvalds 已提交
1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706

/***************************************************************************/
/*                    FUNCTION DECLARATIONS                                */
/***************************************************************************/

#define get_journal_desc_magic(bh) (bh->b_data + bh->b_size - 12)

#define journal_trans_half(blocksize) \
	((blocksize - sizeof (struct reiserfs_journal_desc) + sizeof (__u32) - 12) / sizeof (__u32))

/* journal.c see journal.c for all the comments here */

/* first block written in a commit.  */
struct reiserfs_journal_desc {
1707 1708 1709 1710 1711
	__le32 j_trans_id;	/* id of commit */
	__le32 j_len;		/* length of commit. len +1 is the commit block */
	__le32 j_mount_id;	/* mount id of this trans */
	__le32 j_realblock[1];	/* real locations for each block */
};
L
Linus Torvalds 已提交
1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722

#define get_desc_trans_id(d)   le32_to_cpu((d)->j_trans_id)
#define get_desc_trans_len(d)  le32_to_cpu((d)->j_len)
#define get_desc_mount_id(d)   le32_to_cpu((d)->j_mount_id)

#define set_desc_trans_id(d,val)       do { (d)->j_trans_id = cpu_to_le32 (val); } while (0)
#define set_desc_trans_len(d,val)      do { (d)->j_len = cpu_to_le32 (val); } while (0)
#define set_desc_mount_id(d,val)       do { (d)->j_mount_id = cpu_to_le32 (val); } while (0)

/* last block written in a commit */
struct reiserfs_journal_commit {
1723 1724 1725 1726
	__le32 j_trans_id;	/* must match j_trans_id from the desc block */
	__le32 j_len;		/* ditto */
	__le32 j_realblock[1];	/* real locations for each block */
};
L
Linus Torvalds 已提交
1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739

#define get_commit_trans_id(c) le32_to_cpu((c)->j_trans_id)
#define get_commit_trans_len(c)        le32_to_cpu((c)->j_len)
#define get_commit_mount_id(c) le32_to_cpu((c)->j_mount_id)

#define set_commit_trans_id(c,val)     do { (c)->j_trans_id = cpu_to_le32 (val); } while (0)
#define set_commit_trans_len(c,val)    do { (c)->j_len = cpu_to_le32 (val); } while (0)

/* this header block gets written whenever a transaction is considered fully flushed, and is more recent than the
** last fully flushed transaction.  fully flushed means all the log blocks and all the real blocks are on disk,
** and this transaction does not need to be replayed.
*/
struct reiserfs_journal_header {
1740 1741 1742 1743 1744
	__le32 j_last_flush_trans_id;	/* id of last fully flushed transaction */
	__le32 j_first_unflushed_offset;	/* offset in the log of where to start replay after a crash */
	__le32 j_mount_id;
	/* 12 */ struct journal_params jh_journal;
};
L
Linus Torvalds 已提交
1745 1746

/* biggest tunable defines are right here */
1747 1748
#define JOURNAL_BLOCK_COUNT 8192	/* number of blocks in the journal */
#define JOURNAL_TRANS_MAX_DEFAULT 1024	/* biggest possible single transaction, don't change for now (8/3/99) */
L
Linus Torvalds 已提交
1749
#define JOURNAL_TRANS_MIN_DEFAULT 256
1750
#define JOURNAL_MAX_BATCH_DEFAULT   900	/* max blocks to batch into one transaction, don't make this any bigger than 900 */
L
Linus Torvalds 已提交
1751
#define JOURNAL_MIN_RATIO 2
1752
#define JOURNAL_MAX_COMMIT_AGE 30
L
Linus Torvalds 已提交
1753 1754
#define JOURNAL_MAX_TRANS_AGE 30
#define JOURNAL_PER_BALANCE_CNT (3 * (MAX_HEIGHT-2) + 9)
J
Jeff Mahoney 已提交
1755 1756 1757 1758
#define JOURNAL_BLOCKS_PER_OBJECT(sb)  (JOURNAL_PER_BALANCE_CNT * 3 + \
					 2 * (REISERFS_QUOTA_INIT_BLOCKS(sb) + \
					      REISERFS_QUOTA_TRANS_BLOCKS(sb)))

L
Linus Torvalds 已提交
1759
#ifdef CONFIG_QUOTA
1760 1761 1762 1763 1764 1765 1766 1767
/* We need to update data and inode (atime) */
#define REISERFS_QUOTA_TRANS_BLOCKS(s) (REISERFS_SB(s)->s_mount_opt & (1<<REISERFS_QUOTA) ? 2 : 0)
/* 1 balancing, 1 bitmap, 1 data per write + stat data update */
#define REISERFS_QUOTA_INIT_BLOCKS(s) (REISERFS_SB(s)->s_mount_opt & (1<<REISERFS_QUOTA) ? \
(DQUOT_INIT_ALLOC*(JOURNAL_PER_BALANCE_CNT+2)+DQUOT_INIT_REWRITE+1) : 0)
/* same as with INIT */
#define REISERFS_QUOTA_DEL_BLOCKS(s) (REISERFS_SB(s)->s_mount_opt & (1<<REISERFS_QUOTA) ? \
(DQUOT_DEL_ALLOC*(JOURNAL_PER_BALANCE_CNT+2)+DQUOT_DEL_REWRITE+1) : 0)
L
Linus Torvalds 已提交
1768
#else
1769 1770 1771
#define REISERFS_QUOTA_TRANS_BLOCKS(s) 0
#define REISERFS_QUOTA_INIT_BLOCKS(s) 0
#define REISERFS_QUOTA_DEL_BLOCKS(s) 0
L
Linus Torvalds 已提交
1772 1773 1774 1775 1776 1777 1778 1779
#endif

/* both of these can be as low as 1, or as high as you want.  The min is the
** number of 4k bitmap nodes preallocated on mount. New nodes are allocated
** as needed, and released when transactions are committed.  On release, if 
** the current number of nodes is > max, the node is freed, otherwise, 
** it is put on a free list for faster use later.
*/
1780 1781
#define REISERFS_MIN_BITMAP_NODES 10
#define REISERFS_MAX_BITMAP_NODES 100
L
Linus Torvalds 已提交
1782

1783
#define JBH_HASH_SHIFT 13	/* these are based on journal hash size of 8192 */
L
Linus Torvalds 已提交
1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796
#define JBH_HASH_MASK 8191

#define _jhashfn(sb,block)	\
	(((unsigned long)sb>>L1_CACHE_SHIFT) ^ \
	 (((block)<<(JBH_HASH_SHIFT - 6)) ^ ((block) >> 13) ^ ((block) << (JBH_HASH_SHIFT - 12))))
#define journal_hash(t,sb,block) ((t)[_jhashfn((sb),(block)) & JBH_HASH_MASK])

// We need these to make journal.c code more readable
#define journal_find_get_block(s, block) __find_get_block(SB_JOURNAL(s)->j_dev_bd, block, s->s_blocksize)
#define journal_getblk(s, block) __getblk(SB_JOURNAL(s)->j_dev_bd, block, s->s_blocksize)
#define journal_bread(s, block) __bread(SB_JOURNAL(s)->j_dev_bd, block, s->s_blocksize)

enum reiserfs_bh_state_bits {
1797 1798 1799 1800 1801 1802 1803 1804
	BH_JDirty = BH_PrivateStart,	/* buffer is in current transaction */
	BH_JDirty_wait,
	BH_JNew,		/* disk block was taken off free list before
				 * being in a finished transaction, or
				 * written to disk. Can be reused immed. */
	BH_JPrepared,
	BH_JRestore_dirty,
	BH_JTest,		// debugging only will go away
L
Linus Torvalds 已提交
1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823
};

BUFFER_FNS(JDirty, journaled);
TAS_BUFFER_FNS(JDirty, journaled);
BUFFER_FNS(JDirty_wait, journal_dirty);
TAS_BUFFER_FNS(JDirty_wait, journal_dirty);
BUFFER_FNS(JNew, journal_new);
TAS_BUFFER_FNS(JNew, journal_new);
BUFFER_FNS(JPrepared, journal_prepared);
TAS_BUFFER_FNS(JPrepared, journal_prepared);
BUFFER_FNS(JRestore_dirty, journal_restore_dirty);
TAS_BUFFER_FNS(JRestore_dirty, journal_restore_dirty);
BUFFER_FNS(JTest, journal_test);
TAS_BUFFER_FNS(JTest, journal_test);

/*
** transaction handle which is passed around for all journal calls
*/
struct reiserfs_transaction_handle {
1824 1825 1826 1827 1828 1829 1830 1831 1832
	struct super_block *t_super;	/* super for this FS when journal_begin was
					   called. saves calls to reiserfs_get_super
					   also used by nested transactions to make
					   sure they are nesting on the right FS
					   _must_ be first in the handle
					 */
	int t_refcount;
	int t_blocks_logged;	/* number of blocks this writer has logged */
	int t_blocks_allocated;	/* number of blocks this writer allocated */
1833
	unsigned int t_trans_id;	/* sanity check, equals the current trans id */
1834 1835 1836 1837 1838
	void *t_handle_save;	/* save existing current->journal_info */
	unsigned displace_new_blocks:1;	/* if new block allocation occurres, that block
					   should be displaced from others */
	struct list_head t_list;
};
L
Linus Torvalds 已提交
1839 1840 1841 1842 1843

/* used to keep track of ordered and tail writes, attached to the buffer
 * head through b_journal_head.
 */
struct reiserfs_jh {
1844 1845 1846
	struct reiserfs_journal_list *jl;
	struct buffer_head *bh;
	struct list_head list;
L
Linus Torvalds 已提交
1847 1848 1849 1850 1851
};

void reiserfs_free_jh(struct buffer_head *bh);
int reiserfs_add_tail_list(struct inode *inode, struct buffer_head *bh);
int reiserfs_add_ordered_list(struct inode *inode, struct buffer_head *bh);
1852 1853 1854 1855 1856 1857 1858 1859 1860
int journal_mark_dirty(struct reiserfs_transaction_handle *,
		       struct super_block *, struct buffer_head *bh);

static inline int reiserfs_file_data_log(struct inode *inode)
{
	if (reiserfs_data_log(inode->i_sb) ||
	    (REISERFS_I(inode)->i_flags & i_data_log))
		return 1;
	return 0;
L
Linus Torvalds 已提交
1861 1862
}

1863 1864 1865 1866 1867 1868 1869 1870
static inline int reiserfs_transaction_running(struct super_block *s)
{
	struct reiserfs_transaction_handle *th = current->journal_info;
	if (th && th->t_super == s)
		return 1;
	if (th && th->t_super == NULL)
		BUG();
	return 0;
L
Linus Torvalds 已提交
1871 1872
}

1873 1874 1875 1876 1877
static inline int reiserfs_transaction_free_space(struct reiserfs_transaction_handle *th)
{
	return th->t_blocks_allocated - th->t_blocks_logged;
}

1878 1879 1880 1881
struct reiserfs_transaction_handle *reiserfs_persistent_transaction(struct
								    super_block
								    *,
								    int count);
L
Linus Torvalds 已提交
1882 1883
int reiserfs_end_persistent_transaction(struct reiserfs_transaction_handle *);
int reiserfs_commit_page(struct inode *inode, struct page *page,
1884
			 unsigned from, unsigned to);
L
Linus Torvalds 已提交
1885
int reiserfs_flush_old_commits(struct super_block *);
1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908
int reiserfs_commit_for_inode(struct inode *);
int reiserfs_inode_needs_commit(struct inode *);
void reiserfs_update_inode_transaction(struct inode *);
void reiserfs_wait_on_write_block(struct super_block *s);
void reiserfs_block_writes(struct reiserfs_transaction_handle *th);
void reiserfs_allow_writes(struct super_block *s);
void reiserfs_check_lock_depth(struct super_block *s, char *caller);
int reiserfs_prepare_for_journal(struct super_block *, struct buffer_head *bh,
				 int wait);
void reiserfs_restore_prepared_buffer(struct super_block *,
				      struct buffer_head *bh);
int journal_init(struct super_block *, const char *j_dev_name, int old_format,
		 unsigned int);
int journal_release(struct reiserfs_transaction_handle *, struct super_block *);
int journal_release_error(struct reiserfs_transaction_handle *,
			  struct super_block *);
int journal_end(struct reiserfs_transaction_handle *, struct super_block *,
		unsigned long);
int journal_end_sync(struct reiserfs_transaction_handle *, struct super_block *,
		     unsigned long);
int journal_mark_freed(struct reiserfs_transaction_handle *,
		       struct super_block *, b_blocknr_t blocknr);
int journal_transaction_should_end(struct reiserfs_transaction_handle *, int);
J
Jeff Mahoney 已提交
1909 1910
int reiserfs_in_journal(struct super_block *sb, unsigned int bmap_nr,
			 int bit_nr, int searchall, b_blocknr_t *next);
1911
int journal_begin(struct reiserfs_transaction_handle *,
J
Jeff Mahoney 已提交
1912
		  struct super_block *sb, unsigned long);
1913
int journal_join_abort(struct reiserfs_transaction_handle *,
J
Jeff Mahoney 已提交
1914
		       struct super_block *sb, unsigned long);
J
Jeff Mahoney 已提交
1915
void reiserfs_abort_journal(struct super_block *sb, int errno);
1916 1917
void reiserfs_abort(struct super_block *sb, int errno, const char *fmt, ...);
int reiserfs_allocate_list_bitmaps(struct super_block *s,
1918
				   struct reiserfs_list_bitmap *, unsigned int);
1919 1920 1921 1922

void add_save_link(struct reiserfs_transaction_handle *th,
		   struct inode *inode, int truncate);
int remove_save_link(struct inode *inode, int truncate);
L
Linus Torvalds 已提交
1923 1924

/* objectid.c */
1925 1926 1927 1928
__u32 reiserfs_get_unused_objectid(struct reiserfs_transaction_handle *th);
void reiserfs_release_objectid(struct reiserfs_transaction_handle *th,
			       __u32 objectid_to_release);
int reiserfs_convert_objectid_map_v1(struct super_block *);
L
Linus Torvalds 已提交
1929 1930 1931

/* stree.c */
int B_IS_IN_TREE(const struct buffer_head *);
J
Jeff Mahoney 已提交
1932 1933
extern void copy_item_head(struct item_head *to,
			   const struct item_head *from);
L
Linus Torvalds 已提交
1934 1935

// first key is in cpu form, second - le
1936 1937 1938
extern int comp_short_keys(const struct reiserfs_key *le_key,
			   const struct cpu_key *cpu_key);
extern void le_key2cpu_key(struct cpu_key *to, const struct reiserfs_key *from);
L
Linus Torvalds 已提交
1939 1940

// both are in le form
1941 1942 1943 1944
extern int comp_le_keys(const struct reiserfs_key *,
			const struct reiserfs_key *);
extern int comp_short_le_keys(const struct reiserfs_key *,
			      const struct reiserfs_key *);
L
Linus Torvalds 已提交
1945 1946 1947 1948

//
// get key version from on disk key - kludge
//
1949
static inline int le_key_version(const struct reiserfs_key *key)
L
Linus Torvalds 已提交
1950
{
1951
	int type;
L
Linus Torvalds 已提交
1952

1953 1954 1955 1956 1957 1958
	type = offset_v2_k_type(&(key->u.k_offset_v2));
	if (type != TYPE_DIRECT && type != TYPE_INDIRECT
	    && type != TYPE_DIRENTRY)
		return KEY_FORMAT_3_5;

	return KEY_FORMAT_3_6;
L
Linus Torvalds 已提交
1959 1960 1961

}

1962 1963 1964 1965 1966
static inline void copy_key(struct reiserfs_key *to,
			    const struct reiserfs_key *from)
{
	memcpy(to, from, KEY_SIZE);
}
L
Linus Torvalds 已提交
1967

J
Jeff Mahoney 已提交
1968 1969
int comp_items(const struct item_head *stored_ih, const struct treepath *path);
const struct reiserfs_key *get_rkey(const struct treepath *chk_path,
J
Jeff Mahoney 已提交
1970
				    const struct super_block *sb);
1971
int search_by_key(struct super_block *, const struct cpu_key *,
1972
		  struct treepath *, int);
L
Linus Torvalds 已提交
1973
#define search_item(s,key,path) search_by_key (s, key, path, DISK_LEAF_NODE_LEVEL)
J
Jeff Mahoney 已提交
1974
int search_for_position_by_key(struct super_block *sb,
J
Jeff Mahoney 已提交
1975 1976
			       const struct cpu_key *cpu_key,
			       struct treepath *search_path);
J
Jeff Mahoney 已提交
1977
extern void decrement_bcount(struct buffer_head *bh);
J
Jeff Mahoney 已提交
1978 1979
void decrement_counters_in_path(struct treepath *search_path);
void pathrelse(struct treepath *search_path);
1980
int reiserfs_check_path(struct treepath *p);
J
Jeff Mahoney 已提交
1981
void pathrelse_and_restore(struct super_block *s, struct treepath *search_path);
1982 1983

int reiserfs_insert_item(struct reiserfs_transaction_handle *th,
1984
			 struct treepath *path,
1985 1986 1987 1988 1989
			 const struct cpu_key *key,
			 struct item_head *ih,
			 struct inode *inode, const char *body);

int reiserfs_paste_into_item(struct reiserfs_transaction_handle *th,
1990
			     struct treepath *path,
1991 1992 1993 1994 1995
			     const struct cpu_key *key,
			     struct inode *inode,
			     const char *body, int paste_size);

int reiserfs_cut_from_item(struct reiserfs_transaction_handle *th,
1996
			   struct treepath *path,
1997 1998 1999 2000 2001
			   struct cpu_key *key,
			   struct inode *inode,
			   struct page *page, loff_t new_file_size);

int reiserfs_delete_item(struct reiserfs_transaction_handle *th,
2002
			 struct treepath *path,
2003
			 const struct cpu_key *key,
J
Jeff Mahoney 已提交
2004
			 struct inode *inode, struct buffer_head *un_bh);
2005 2006 2007 2008

void reiserfs_delete_solid_item(struct reiserfs_transaction_handle *th,
				struct inode *inode, struct reiserfs_key *key);
int reiserfs_delete_object(struct reiserfs_transaction_handle *th,
2009
			   struct inode *inode);
2010
int reiserfs_do_truncate(struct reiserfs_transaction_handle *th,
2011
			 struct inode *inode, struct page *,
2012
			 int update_timestamps);
L
Linus Torvalds 已提交
2013 2014 2015 2016 2017 2018 2019 2020

#define i_block_size(inode) ((inode)->i_sb->s_blocksize)
#define file_size(inode) ((inode)->i_size)
#define tail_size(inode) (file_size (inode) & (i_block_size (inode) - 1))

#define tail_has_to_be_packed(inode) (have_large_tails ((inode)->i_sb)?\
!STORE_TAIL_IN_UNFM_S1(file_size (inode), tail_size(inode), inode->i_sb->s_blocksize):have_small_tails ((inode)->i_sb)?!STORE_TAIL_IN_UNFM_S2(file_size (inode), tail_size(inode), inode->i_sb->s_blocksize):0 )

2021
void padd_item(char *item, int total_length, int length);
L
Linus Torvalds 已提交
2022 2023 2024

/* inode.c */
/* args for the create parameter of reiserfs_get_block */
2025 2026 2027 2028
#define GET_BLOCK_NO_CREATE 0	/* don't create new blocks or convert tails */
#define GET_BLOCK_CREATE 1	/* add anything you need to find block */
#define GET_BLOCK_NO_HOLE 2	/* return -ENOENT for file holes */
#define GET_BLOCK_READ_DIRECT 4	/* read the tail if indirect item not found */
2029
#define GET_BLOCK_NO_IMUX     8	/* i_mutex is not held, don't preallocate */
2030 2031 2032 2033 2034 2035 2036
#define GET_BLOCK_NO_DANGLE   16	/* don't leave any transactions running */

void reiserfs_read_locked_inode(struct inode *inode,
				struct reiserfs_iget_args *args);
int reiserfs_find_actor(struct inode *inode, void *p);
int reiserfs_init_locked_inode(struct inode *inode, void *p);
void reiserfs_delete_inode(struct inode *inode);
2037
int reiserfs_write_inode(struct inode *inode, struct writeback_control *wbc);
2038 2039
int reiserfs_get_block(struct inode *inode, sector_t block,
		       struct buffer_head *bh_result, int create);
C
Christoph Hellwig 已提交
2040 2041 2042 2043
struct dentry *reiserfs_fh_to_dentry(struct super_block *sb, struct fid *fid,
				     int fh_len, int fh_type);
struct dentry *reiserfs_fh_to_parent(struct super_block *sb, struct fid *fid,
				     int fh_len, int fh_type);
2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054
int reiserfs_encode_fh(struct dentry *dentry, __u32 * data, int *lenp,
		       int connectable);

int reiserfs_truncate_file(struct inode *, int update_timestamps);
void make_cpu_key(struct cpu_key *cpu_key, struct inode *inode, loff_t offset,
		  int type, int key_length);
void make_le_item_head(struct item_head *ih, const struct cpu_key *key,
		       int version,
		       loff_t offset, int type, int length, int entry_count);
struct inode *reiserfs_iget(struct super_block *s, const struct cpu_key *key);

2055
struct reiserfs_security_handle;
2056 2057 2058
int reiserfs_new_inode(struct reiserfs_transaction_handle *th,
		       struct inode *dir, int mode,
		       const char *symname, loff_t i_size,
2059 2060
		       struct dentry *dentry, struct inode *inode,
		       struct reiserfs_security_handle *security);
2061 2062 2063

void reiserfs_update_sd_size(struct reiserfs_transaction_handle *th,
			     struct inode *inode, loff_t size);
L
Linus Torvalds 已提交
2064 2065

static inline void reiserfs_update_sd(struct reiserfs_transaction_handle *th,
2066
				      struct inode *inode)
L
Linus Torvalds 已提交
2067
{
2068
	reiserfs_update_sd_size(th, inode, inode->i_size);
L
Linus Torvalds 已提交
2069 2070
}

2071 2072
void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode);
void i_attrs_to_sd_attrs(struct inode *inode, __u16 * sd_attrs);
L
Linus Torvalds 已提交
2073 2074 2075
int reiserfs_setattr(struct dentry *dentry, struct iattr *attr);

/* namei.c */
2076 2077
void set_de_name_and_namelen(struct reiserfs_dir_entry *de);
int search_by_entry_key(struct super_block *sb, const struct cpu_key *key,
2078
			struct treepath *path, struct reiserfs_dir_entry *de);
2079
struct dentry *reiserfs_get_parent(struct dentry *);
L
Linus Torvalds 已提交
2080

2081
#ifdef CONFIG_REISERFS_PROC_INFO
2082 2083 2084 2085
int reiserfs_proc_info_init(struct super_block *sb);
int reiserfs_proc_info_done(struct super_block *sb);
int reiserfs_proc_info_global_init(void);
int reiserfs_proc_info_global_done(void);
L
Linus Torvalds 已提交
2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099

#define PROC_EXP( e )   e

#define __PINFO( sb ) REISERFS_SB(sb) -> s_proc_info_data
#define PROC_INFO_MAX( sb, field, value )								\
    __PINFO( sb ).field =												\
        max( REISERFS_SB( sb ) -> s_proc_info_data.field, value )
#define PROC_INFO_INC( sb, field ) ( ++ ( __PINFO( sb ).field ) )
#define PROC_INFO_ADD( sb, field, val ) ( __PINFO( sb ).field += ( val ) )
#define PROC_INFO_BH_STAT( sb, bh, level )							\
    PROC_INFO_INC( sb, sbk_read_at[ ( level ) ] );						\
    PROC_INFO_ADD( sb, free_at[ ( level ) ], B_FREE_SPACE( bh ) );	\
    PROC_INFO_ADD( sb, items_at[ ( level ) ], B_NR_ITEMS( bh ) )
#else
2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119
static inline int reiserfs_proc_info_init(struct super_block *sb)
{
	return 0;
}

static inline int reiserfs_proc_info_done(struct super_block *sb)
{
	return 0;
}

static inline int reiserfs_proc_info_global_init(void)
{
	return 0;
}

static inline int reiserfs_proc_info_global_done(void)
{
	return 0;
}

L
Linus Torvalds 已提交
2120 2121 2122 2123 2124
#define PROC_EXP( e )
#define VOID_V ( ( void ) 0 )
#define PROC_INFO_MAX( sb, field, value ) VOID_V
#define PROC_INFO_INC( sb, field ) VOID_V
#define PROC_INFO_ADD( sb, field, val ) VOID_V
J
Jeff Mahoney 已提交
2125
#define PROC_INFO_BH_STAT(sb, bh, n_node_level) VOID_V
L
Linus Torvalds 已提交
2126 2127 2128
#endif

/* dir.c */
2129 2130 2131
extern const struct inode_operations reiserfs_dir_inode_operations;
extern const struct inode_operations reiserfs_symlink_inode_operations;
extern const struct inode_operations reiserfs_special_inode_operations;
2132
extern const struct file_operations reiserfs_dir_operations;
2133
int reiserfs_readdir_dentry(struct dentry *, void *, filldir_t, loff_t *);
L
Linus Torvalds 已提交
2134 2135

/* tail_conversion.c */
2136
int direct2indirect(struct reiserfs_transaction_handle *, struct inode *,
2137
		    struct treepath *, struct buffer_head *, loff_t);
2138
int indirect2direct(struct reiserfs_transaction_handle *, struct inode *,
2139
		    struct page *, struct treepath *, const struct cpu_key *,
2140 2141
		    loff_t, char *);
void reiserfs_unmap_buffer(struct buffer_head *);
L
Linus Torvalds 已提交
2142 2143

/* file.c */
2144
extern const struct inode_operations reiserfs_file_inode_operations;
2145
extern const struct file_operations reiserfs_file_operations;
2146
extern const struct address_space_operations reiserfs_address_space_operations;
L
Linus Torvalds 已提交
2147 2148 2149

/* fix_nodes.c */

J
Jeff Mahoney 已提交
2150
int fix_nodes(int n_op_mode, struct tree_balance *tb,
J
Jeff Mahoney 已提交
2151
	      struct item_head *ins_ih, const void *);
2152
void unfix_nodes(struct tree_balance *);
L
Linus Torvalds 已提交
2153 2154

/* prints.c */
J
Jeff Mahoney 已提交
2155 2156
void __reiserfs_panic(struct super_block *s, const char *id,
		      const char *function, const char *fmt, ...)
2157
    __attribute__ ((noreturn));
J
Jeff Mahoney 已提交
2158 2159
#define reiserfs_panic(s, id, fmt, args...) \
	__reiserfs_panic(s, id, __func__, fmt, ##args)
2160 2161 2162 2163
void __reiserfs_error(struct super_block *s, const char *id,
		      const char *function, const char *fmt, ...);
#define reiserfs_error(s, id, fmt, args...) \
	 __reiserfs_error(s, id, __func__, fmt, ##args)
2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176
void reiserfs_info(struct super_block *s, const char *fmt, ...);
void reiserfs_debug(struct super_block *s, int level, const char *fmt, ...);
void print_indirect_item(struct buffer_head *bh, int item_num);
void store_print_tb(struct tree_balance *tb);
void print_cur_tb(char *mes);
void print_de(struct reiserfs_dir_entry *de);
void print_bi(struct buffer_info *bi, char *mes);
#define PRINT_LEAF_ITEMS 1	/* print all items */
#define PRINT_DIRECTORY_ITEMS 2	/* print directory items */
#define PRINT_DIRECT_ITEMS 4	/* print contents of direct items */
void print_block(struct buffer_head *bh, ...);
void print_bmap(struct super_block *s, int silent);
void print_bmap_block(int i, char *data, int size, int silent);
L
Linus Torvalds 已提交
2177
/*void print_super_block (struct super_block * s, char * mes);*/
2178 2179 2180 2181 2182 2183
void print_objectid_map(struct super_block *s);
void print_block_head(struct buffer_head *bh, char *mes);
void check_leaf(struct buffer_head *bh);
void check_internal(struct buffer_head *bh);
void print_statistics(struct super_block *s);
char *reiserfs_hashname(int code);
L
Linus Torvalds 已提交
2184 2185

/* lbalance.c */
2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199
int leaf_move_items(int shift_mode, struct tree_balance *tb, int mov_num,
		    int mov_bytes, struct buffer_head *Snew);
int leaf_shift_left(struct tree_balance *tb, int shift_num, int shift_bytes);
int leaf_shift_right(struct tree_balance *tb, int shift_num, int shift_bytes);
void leaf_delete_items(struct buffer_info *cur_bi, int last_first, int first,
		       int del_num, int del_bytes);
void leaf_insert_into_buf(struct buffer_info *bi, int before,
			  struct item_head *inserted_item_ih,
			  const char *inserted_item_body, int zeros_number);
void leaf_paste_in_buffer(struct buffer_info *bi, int pasted_item_num,
			  int pos_in_item, int paste_size, const char *body,
			  int zeros_number);
void leaf_cut_from_buffer(struct buffer_info *bi, int cut_item_num,
			  int pos_in_item, int cut_size);
2200
void leaf_paste_entries(struct buffer_info *bi, int item_num, int before,
2201 2202
			int new_entry_count, struct reiserfs_de_head *new_dehs,
			const char *records, int paste_size);
L
Linus Torvalds 已提交
2203
/* ibalance.c */
2204 2205
int balance_internal(struct tree_balance *, int, int, struct item_head *,
		     struct buffer_head **);
L
Linus Torvalds 已提交
2206 2207

/* do_balance.c */
2208 2209
void do_balance_mark_leaf_dirty(struct tree_balance *tb,
				struct buffer_head *bh, int flag);
L
Linus Torvalds 已提交
2210 2211 2212
#define do_balance_mark_internal_dirty do_balance_mark_leaf_dirty
#define do_balance_mark_sb_dirty do_balance_mark_leaf_dirty

2213 2214 2215 2216
void do_balance(struct tree_balance *tb, struct item_head *ih,
		const char *body, int flag);
void reiserfs_invalidate_buffer(struct tree_balance *tb,
				struct buffer_head *bh);
L
Linus Torvalds 已提交
2217

2218 2219 2220 2221 2222 2223
int get_left_neighbor_position(struct tree_balance *tb, int h);
int get_right_neighbor_position(struct tree_balance *tb, int h);
void replace_key(struct tree_balance *tb, struct buffer_head *, int,
		 struct buffer_head *, int);
void make_empty_node(struct buffer_info *);
struct buffer_head *get_FEB(struct tree_balance *);
L
Linus Torvalds 已提交
2224 2225 2226 2227 2228

/* bitmap.c */

/* structure contains hints for block allocator, and it is a container for
 * arguments, such as node, search path, transaction_handle, etc. */
2229 2230
struct __reiserfs_blocknr_hint {
	struct inode *inode;	/* inode passed to allocator, if we allocate unf. nodes */
2231
	sector_t block;		/* file offset, in blocks */
2232
	struct in_core_key key;
2233
	struct treepath *path;	/* search path, used by allocator to deternine search_start by
2234 2235 2236 2237 2238
				 * various ways */
	struct reiserfs_transaction_handle *th;	/* transaction handle is needed to log super blocks and
						 * bitmap blocks changes  */
	b_blocknr_t beg, end;
	b_blocknr_t search_start;	/* a field used to transfer search start value (block number)
L
Linus Torvalds 已提交
2239 2240
					 * between different block allocator procedures
					 * (determine_search_start() and others) */
2241 2242
	int prealloc_size;	/* is set in determine_prealloc_size() function, used by underlayed
				 * function that do actual allocation */
L
Linus Torvalds 已提交
2243

2244
	unsigned formatted_node:1;	/* the allocator uses different polices for getting disk space for
L
Linus Torvalds 已提交
2245
					 * formatted/unformatted blocks with/without preallocation */
2246
	unsigned preallocate:1;
L
Linus Torvalds 已提交
2247 2248 2249 2250
};

typedef struct __reiserfs_blocknr_hint reiserfs_blocknr_hint_t;

2251 2252
int reiserfs_parse_alloc_options(struct super_block *, char *);
void reiserfs_init_alloc_options(struct super_block *s);
L
Linus Torvalds 已提交
2253 2254 2255 2256 2257 2258

/*
 * given a directory, this will tell you what packing locality
 * to use for a new object underneat it.  The locality is returned
 * in disk byte order (le).
 */
2259
__le32 reiserfs_choose_packing(struct inode *dir);
L
Linus Torvalds 已提交
2260

2261 2262 2263 2264
int reiserfs_init_bitmap_cache(struct super_block *sb);
void reiserfs_free_bitmap_cache(struct super_block *sb);
void reiserfs_cache_bitmap_metadata(struct super_block *sb, struct buffer_head *bh, struct reiserfs_bitmap_info *info);
struct buffer_head *reiserfs_read_bitmap_block(struct super_block *sb, unsigned int bitmap);
2265 2266 2267 2268 2269
int is_reusable(struct super_block *s, b_blocknr_t block, int bit_value);
void reiserfs_free_block(struct reiserfs_transaction_handle *th, struct inode *,
			 b_blocknr_t, int for_unformatted);
int reiserfs_allocate_blocknrs(reiserfs_blocknr_hint_t *, b_blocknr_t *, int,
			       int);
2270
static inline int reiserfs_new_form_blocknrs(struct tree_balance *tb,
2271 2272
					     b_blocknr_t * new_blocknrs,
					     int amount_needed)
L
Linus Torvalds 已提交
2273
{
2274 2275 2276 2277 2278 2279 2280 2281 2282 2283
	reiserfs_blocknr_hint_t hint = {
		.th = tb->transaction_handle,
		.path = tb->tb_path,
		.inode = NULL,
		.key = tb->key,
		.block = 0,
		.formatted_node = 1
	};
	return reiserfs_allocate_blocknrs(&hint, new_blocknrs, amount_needed,
					  0);
L
Linus Torvalds 已提交
2284 2285
}

2286
static inline int reiserfs_new_unf_blocknrs(struct reiserfs_transaction_handle
2287 2288
					    *th, struct inode *inode,
					    b_blocknr_t * new_blocknrs,
2289 2290
					    struct treepath *path,
					    sector_t block)
L
Linus Torvalds 已提交
2291
{
2292 2293 2294 2295 2296 2297 2298 2299 2300
	reiserfs_blocknr_hint_t hint = {
		.th = th,
		.path = path,
		.inode = inode,
		.block = block,
		.formatted_node = 0,
		.preallocate = 0
	};
	return reiserfs_allocate_blocknrs(&hint, new_blocknrs, 1, 0);
L
Linus Torvalds 已提交
2301 2302 2303
}

#ifdef REISERFS_PREALLOCATE
2304
static inline int reiserfs_new_unf_blocknrs2(struct reiserfs_transaction_handle
2305 2306
					     *th, struct inode *inode,
					     b_blocknr_t * new_blocknrs,
2307 2308
					     struct treepath *path,
					     sector_t block)
L
Linus Torvalds 已提交
2309
{
2310 2311 2312 2313 2314 2315 2316 2317 2318
	reiserfs_blocknr_hint_t hint = {
		.th = th,
		.path = path,
		.inode = inode,
		.block = block,
		.formatted_node = 0,
		.preallocate = 1
	};
	return reiserfs_allocate_blocknrs(&hint, new_blocknrs, 1, 0);
L
Linus Torvalds 已提交
2319 2320
}

2321 2322 2323
void reiserfs_discard_prealloc(struct reiserfs_transaction_handle *th,
			       struct inode *inode);
void reiserfs_discard_all_prealloc(struct reiserfs_transaction_handle *th);
L
Linus Torvalds 已提交
2324 2325 2326
#endif

/* hashes.c */
2327 2328 2329
__u32 keyed_hash(const signed char *msg, int len);
__u32 yura_hash(const signed char *msg, int len);
__u32 r5_hash(const signed char *msg, int len);
L
Linus Torvalds 已提交
2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349

/* the ext2 bit routines adjust for big or little endian as
** appropriate for the arch, so in our laziness we use them rather
** than using the bit routines they call more directly.  These
** routines must be used when changing on disk bitmaps.  */
#define reiserfs_test_and_set_le_bit   ext2_set_bit
#define reiserfs_test_and_clear_le_bit ext2_clear_bit
#define reiserfs_test_le_bit           ext2_test_bit
#define reiserfs_find_next_zero_le_bit ext2_find_next_zero_bit

/* sometimes reiserfs_truncate may require to allocate few new blocks
   to perform indirect2direct conversion. People probably used to
   think, that truncate should work without problems on a filesystem
   without free disk space. They may complain that they can not
   truncate due to lack of free disk space. This spare space allows us
   to not worry about it. 500 is probably too much, but it should be
   absolutely safe */
#define SPARE_SPACE 500

/* prototypes from ioctl.c */
2350
long reiserfs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
2351 2352
long reiserfs_compat_ioctl(struct file *filp,
		   unsigned int cmd, unsigned long arg);
J
Jan Kara 已提交
2353
int reiserfs_unpack(struct inode *inode, struct file *filp);
2354

2355
#endif /* __KERNEL__ */
2356 2357

#endif				/* _LINUX_REISER_FS_H */