Subversion Repositories shark

Rev

Rev 422 | Blame | Compare with Previous | Last modification | View Log | RSS feed

/*
 * linux/include/linux/jbd.h
 *
 * Written by Stephen C. Tweedie <sct@redhat.com>
 *
 * Copyright 1998-2000 Red Hat, Inc --- All Rights Reserved
 *
 * This file is part of the Linux kernel and is made available under
 * the terms of the GNU General Public License, version 2, or at your
 * option, any later version, incorporated herein by reference.
 *
 * Definitions for transaction data structures for the buffer cache
 * filesystem journaling support.
 */


#ifndef _LINUX_JBD_H
#define _LINUX_JBD_H

#if defined(CONFIG_JBD) || defined(CONFIG_JBD_MODULE) || !defined(__KERNEL__)

/* Allow this file to be included directly into e2fsprogs */
#ifndef __KERNEL__
#include "jfs_compat.h"
#define JFS_DEBUG
#define jfs_debug jbd_debug
#else

#include <linux/buffer_head.h>
#include <linux/journal-head.h>
#include <linux/stddef.h>
#include <asm/semaphore.h>
#endif

#define journal_oom_retry 1

/*
 * Define JBD_PARANIOD_IOFAIL to cause a kernel BUG() if ext3 finds
 * certain classes of error which can occur due to failed IOs.  Under
 * normal use we want ext3 to continue after such errors, because
 * hardware _can_ fail, but for debugging purposes when running tests on
 * known-good hardware we may want to trap these errors.
 */

#undef JBD_PARANOID_IOFAIL

#ifdef CONFIG_JBD_DEBUG
/*
 * Define JBD_EXPENSIVE_CHECKING to enable more expensive internal
 * consistency checks.  By default we don't do this unless
 * CONFIG_JBD_DEBUG is on.
 */

#define JBD_EXPENSIVE_CHECKING
extern int journal_enable_debug;

#define jbd_debug(n, f, a...)                                           \
        do {                                                            \
                if ((n) <= journal_enable_debug) {                      \
                        printk (KERN_DEBUG "(%s, %d): %s: ",            \
                                __FILE__, __LINE__, __FUNCTION__);      \
                        printk (f, ## a);                               \
                }                                                       \
        } while (0)

#else
#define jbd_debug(f, a...)      /**/
#endif

extern void * __jbd_kmalloc (const char *where, size_t size, int flags, int retry);
#define jbd_kmalloc(size, flags) \
        __jbd_kmalloc(__FUNCTION__, (size), (flags), journal_oom_retry)

#define jbd_rep_kmalloc(size, flags) \
        __jbd_kmalloc(__FUNCTION__, (size), (flags), 1)


#define JFS_MIN_JOURNAL_BLOCKS 1024

#ifdef __KERNEL__

/**
 * typedef handle_t - The handle_t type represents a single atomic update being performed by some process.
 *
 * All filesystem modifications made by the process go
 * through this handle.  Recursive operations (such as quota operations)
 * are gathered into a single update.
 *
 * The buffer credits field is used to account for journaled buffers
 * being modified by the running process.  To ensure that there is
 * enough log space for all outstanding operations, we need to limit the
 * number of outstanding buffers possible at any time.  When the
 * operation completes, any buffer credits not used are credited back to
 * the transaction, so that at all times we know how many buffers the
 * outstanding updates on a transaction might possibly touch.
 *
 * This is an opaque datatype.
 **/

typedef struct handle_s         handle_t;       /* Atomic operation type */


/**
 * typedef journal_t - The journal_t maintains all of the journaling state information for a single filesystem.
 *
 * journal_t is linked to from the fs superblock structure.
 *
 * We use the journal_t to keep track of all outstanding transaction
 * activity on the filesystem, and to manage the state of the log
 * writing process.
 *
 * This is an opaque datatype.
 **/

typedef struct journal_s        journal_t;      /* Journal control structure */
#endif

/*
 * Internal structures used by the logging mechanism:
 */


#define JFS_MAGIC_NUMBER 0xc03b3998U /* The first 4 bytes of /dev/random! */

/*
 * On-disk structures
 */


/*
 * Descriptor block types:
 */


#define JFS_DESCRIPTOR_BLOCK    1
#define JFS_COMMIT_BLOCK        2
#define JFS_SUPERBLOCK_V1       3
#define JFS_SUPERBLOCK_V2       4
#define JFS_REVOKE_BLOCK        5

/*
 * Standard header for all descriptor blocks:
 */

typedef struct journal_header_s
{
        __u32           h_magic;
        __u32           h_blocktype;
        __u32           h_sequence;
} journal_header_t;


/*
 * The block tag: used to describe a single buffer in the journal
 */

typedef struct journal_block_tag_s
{
        __u32           t_blocknr;      /* The on-disk block number */
        __u32           t_flags;        /* See below */
} journal_block_tag_t;

/*
 * The revoke descriptor: used on disk to describe a series of blocks to
 * be revoked from the log
 */

typedef struct journal_revoke_header_s
{
        journal_header_t r_header;
        int              r_count;       /* Count of bytes used in the block */
} journal_revoke_header_t;


/* Definitions for the journal tag flags word: */
#define JFS_FLAG_ESCAPE         1       /* on-disk block is escaped */
#define JFS_FLAG_SAME_UUID      2       /* block has same uuid as previous */
#define JFS_FLAG_DELETED        4       /* block deleted by this transaction */
#define JFS_FLAG_LAST_TAG       8       /* last tag in this descriptor block */


/*
 * The journal superblock.  All fields are in big-endian byte order.
 */

typedef struct journal_superblock_s
{
/* 0x0000 */
        journal_header_t s_header;

/* 0x000C */
        /* Static information describing the journal */
        __u32   s_blocksize;            /* journal device blocksize */
        __u32   s_maxlen;               /* total blocks in journal file */
        __u32   s_first;                /* first block of log information */

/* 0x0018 */
        /* Dynamic information describing the current state of the log */
        __u32   s_sequence;             /* first commit ID expected in log */
        __u32   s_start;                /* blocknr of start of log */

/* 0x0020 */
        /* Error value, as set by journal_abort(). */
        __s32   s_errno;

/* 0x0024 */
        /* Remaining fields are only valid in a version-2 superblock */
        __u32   s_feature_compat;       /* compatible feature set */
        __u32   s_feature_incompat;     /* incompatible feature set */
        __u32   s_feature_ro_compat;    /* readonly-compatible feature set */
/* 0x0030 */
        __u8    s_uuid[16];             /* 128-bit uuid for journal */

/* 0x0040 */
        __u32   s_nr_users;             /* Nr of filesystems sharing log */

        __u32   s_dynsuper;             /* Blocknr of dynamic superblock copy*/

/* 0x0048 */
        __u32   s_max_transaction;      /* Limit of journal blocks per trans.*/
        __u32   s_max_trans_data;       /* Limit of data blocks per trans. */

/* 0x0050 */
        __u32   s_padding[44];

/* 0x0100 */
        __u8    s_users[16*48];         /* ids of all fs'es sharing the log */
/* 0x0400 */
} journal_superblock_t;

#define JFS_HAS_COMPAT_FEATURE(j,mask)                                  \
        ((j)->j_format_version >= 2 &&                                  \
         ((j)->j_superblock->s_feature_compat & cpu_to_be32((mask))))

#define JFS_HAS_RO_COMPAT_FEATURE(j,mask)                               \
        ((j)->j_format_version >= 2 &&                                  \
         ((j)->j_superblock->s_feature_ro_compat & cpu_to_be32((mask))))

#define JFS_HAS_INCOMPAT_FEATURE(j,mask)                                \
        ((j)->j_format_version >= 2 &&                                  \
         ((j)->j_superblock->s_feature_incompat & cpu_to_be32((mask))))


#define JFS_FEATURE_INCOMPAT_REVOKE     0x00000001

/* Features known to this kernel version: */
#define JFS_KNOWN_COMPAT_FEATURES       0
#define JFS_KNOWN_ROCOMPAT_FEATURES     0
#define JFS_KNOWN_INCOMPAT_FEATURES     JFS_FEATURE_INCOMPAT_REVOKE

#ifdef __KERNEL__

#include <linux/fs.h>
#include <linux/sched.h>
#include <asm/bug.h>

#define JBD_ASSERTIONS
#ifdef JBD_ASSERTIONS
#define J_ASSERT(assert)                                                \
do {                                                                    \
        if (!(assert)) {                                                \
                printk (KERN_EMERG                                      \
                        "Assertion failure in %s() at %s:%d: \"%s\"\n", \
                        __FUNCTION__, __FILE__, __LINE__, # assert);    \
                BUG();                                                  \
        }                                                               \
} while (0)


#if defined(CONFIG_BUFFER_DEBUG)
void buffer_assertion_failure(struct buffer_head *bh);
#define J_ASSERT_BH(bh, expr)                                           \
        do {                                                            \
                if (!(expr))                                            \
                        buffer_assertion_failure(bh);                   \
                J_ASSERT(expr);                                         \
        } while (0)

#define J_ASSERT_JH(jh, expr)   J_ASSERT_BH(jh2bh(jh), expr)
#else
#define J_ASSERT_BH(bh, expr)   J_ASSERT(expr)
#define J_ASSERT_JH(jh, expr)   J_ASSERT(expr)
#endif

#else
#define J_ASSERT(assert)        do { } while (0)
#endif          /* JBD_ASSERTIONS */

#if defined(JBD_PARANOID_IOFAIL)
#define J_EXPECT(expr, why...)          J_ASSERT(expr)
#define J_EXPECT_BH(bh, expr, why...)   J_ASSERT_BH(bh, expr)
#define J_EXPECT_JH(jh, expr, why...)   J_ASSERT_JH(jh, expr)
#else
#define __journal_expect(expr, why...)                                       \
        do {                                                                 \
                if (!(expr)) {                                               \
                        printk(KERN_ERR                                      \
                                "EXT3-fs unexpected failure: %s;\n",# expr); \
                        printk(KERN_ERR why);                                \
                }                                                            \
        } while (0)

#define J_EXPECT(expr, why...)          __journal_expect(expr, ## why)
#define J_EXPECT_BH(bh, expr, why...)   __journal_expect(expr, ## why)
#define J_EXPECT_JH(jh, expr, why...)   __journal_expect(expr, ## why)
#endif

enum jbd_state_bits {
        BH_JBD                  /* Has an attached ext3 journal_head */
          = BH_PrivateStart,
        BH_JWrite,              /* Being written to log (@@@ DEBUGGING) */
        BH_Freed,               /* Has been freed (truncated) */
        BH_Revoked,             /* Has been revoked from the log */
        BH_RevokeValid,         /* Revoked flag is valid */
        BH_JBDDirty,            /* Is dirty but journaled */
        BH_State,               /* Pins most journal_head state */
        BH_JournalHead,         /* Pins bh->b_private and jh->b_bh */
};

BUFFER_FNS(JBD, jbd)
BUFFER_FNS(JWrite, jwrite)
BUFFER_FNS(JBDDirty, jbddirty)
TAS_BUFFER_FNS(JBDDirty, jbddirty)
BUFFER_FNS(Freed, freed)

static inline struct buffer_head *jh2bh(struct journal_head *jh)
{
        return jh->b_bh;
}

static inline struct journal_head *bh2jh(struct buffer_head *bh)
{
        return bh->b_private;
}

static inline void jbd_lock_bh_state(struct buffer_head *bh)
{
        bit_spin_lock(BH_State, &bh->b_state);
}

static inline int jbd_trylock_bh_state(struct buffer_head *bh)
{
        return bit_spin_trylock(BH_State, &bh->b_state);
}

static inline int jbd_is_locked_bh_state(struct buffer_head *bh)
{
        return bit_spin_is_locked(BH_State, &bh->b_state);
}

static inline void jbd_unlock_bh_state(struct buffer_head *bh)
{
        bit_spin_unlock(BH_State, &bh->b_state);
}

static inline void jbd_lock_bh_journal_head(struct buffer_head *bh)
{
        bit_spin_lock(BH_JournalHead, &bh->b_state);
}

static inline void jbd_unlock_bh_journal_head(struct buffer_head *bh)
{
        bit_spin_unlock(BH_JournalHead, &bh->b_state);
}

#define HAVE_JOURNAL_CALLBACK_STATUS
/**
 *   struct journal_callback - Base structure for callback information.
 *   @jcb_list: list information for other callbacks attached to the same handle.
 *   @jcb_func: Function to call with this callback structure.
 *
 *   This struct is a 'seed' structure for a using with your own callback
 *   structs. If you are using callbacks you must allocate one of these
 *   or another struct of your own definition which has this struct
 *   as it's first element and pass it to journal_callback_set().
 *
 *   This is used internally by jbd to maintain callback information.
 *
 *   See journal_callback_set for more information.
 **/

struct journal_callback {
        struct list_head jcb_list;              /* t_jcb_lock */
        void (*jcb_func)(struct journal_callback *jcb, int error);
        /* user data goes here */
};

struct jbd_revoke_table_s;

/**
 * struct handle_s - The handle_s type is the concrete type associated with
 *     handle_t.
 * @h_transaction: Which compound transaction is this update a part of?
 * @h_buffer_credits: Number of remaining buffers we are allowed to dirty.
 * @h_ref: Reference count on this handle
 * @h_jcb: List of application registered callbacks for this handle.
 * @h_err: Field for caller's use to track errors through large fs operations
 * @h_sync: flag for sync-on-close
 * @h_jdata: flag to force data journaling
 * @h_aborted: flag indicating fatal error on handle
 **/


/* Docbook can't yet cope with the bit fields, but will leave the documentation
 * in so it can be fixed later.
 */


struct handle_s
{
        /* Which compound transaction is this update a part of? */
        transaction_t           *h_transaction;

        /* Number of remaining buffers we are allowed to dirty: */
        int                     h_buffer_credits;

        /* Reference count on this handle */
        int                     h_ref;

        /* Field for caller's use to track errors through large fs */
        /* operations */
        int                     h_err;

        /*
         * List of application registered callbacks for this handle. The
         * function(s) will be called after the transaction that this handle is
         * part of has been committed to disk. [t_jcb_lock]
         */

        struct list_head        h_jcb;

        /* Flags [no locking] */
        unsigned int    h_sync:         1;      /* sync-on-close */
        unsigned int    h_jdata:        1;      /* force data journaling */
        unsigned int    h_aborted:      1;      /* fatal error on handle */
};


/* The transaction_t type is the guts of the journaling mechanism.  It
 * tracks a compound transaction through its various states:
 *
 * RUNNING:     accepting new updates
 * LOCKED:      Updates still running but we don't accept new ones
 * RUNDOWN:     Updates are tidying up but have finished requesting
 *              new buffers to modify (state not used for now)
 * FLUSH:       All updates complete, but we are still writing to disk
 * COMMIT:      All data on disk, writing commit record
 * FINISHED:    We still have to keep the transaction for checkpointing.
 *
 * The transaction keeps track of all of the buffers modified by a
 * running transaction, and all of the buffers committed but not yet
 * flushed to home for finished transactions.
 */


/*
 * Lock ranking:
 *
 *    j_list_lock
 *      ->jbd_lock_bh_journal_head()    (This is "innermost")
 *
 *    j_state_lock
 *    ->jbd_lock_bh_state()
 *
 *    jbd_lock_bh_state()
 *    ->j_list_lock
 *
 *    j_state_lock
 *    ->t_handle_lock
 *
 *    j_state_lock
 *    ->j_list_lock                     (journal_unmap_buffer)
 *
 *    t_handle_lock
 *    ->t_jcb_lock
 */


struct transaction_s
{
        /* Pointer to the journal for this transaction. [no locking] */
        journal_t               *t_journal;

        /* Sequence number for this transaction [no locking] */
        tid_t                   t_tid;

        /*
         * Transaction's current state
         * [no locking - only kjournald alters this]
         * FIXME: needs barriers
         * KLUDGE: [use j_state_lock]
         */

        enum {
                T_RUNNING,
                T_LOCKED,
                T_RUNDOWN,
                T_FLUSH,
                T_COMMIT,
                T_FINISHED
        }                       t_state;

        /*
         * Where in the log does this transaction's commit start? [no locking]
         */

        unsigned long           t_log_start;

        /* Number of buffers on the t_buffers list [j_list_lock] */
        int                     t_nr_buffers;

        /*
         * Doubly-linked circular list of all buffers reserved but not yet
         * modified by this transaction [j_list_lock]
         */

        struct journal_head     *t_reserved_list;

        /*
         * Doubly-linked circular list of all metadata buffers owned by this
         * transaction [j_list_lock]
         */

        struct journal_head     *t_buffers;

        /*
         * Doubly-linked circular list of all data buffers still to be
         * flushed before this transaction can be committed [j_list_lock]
         */

        struct journal_head     *t_sync_datalist;

        /*
         * Doubly-linked circular list of all forget buffers (superseded
         * buffers which we can un-checkpoint once this transaction commits)
         * [j_list_lock]
         */

        struct journal_head     *t_forget;

        /*
         * Doubly-linked circular list of all buffers still to be flushed before
         * this transaction can be checkpointed. [j_list_lock]
         */

        struct journal_head     *t_checkpoint_list;

        /*
         * Doubly-linked circular list of temporary buffers currently undergoing
         * IO in the log [j_list_lock]
         */

        struct journal_head     *t_iobuf_list;

        /*
         * Doubly-linked circular list of metadata buffers being shadowed by log
         * IO.  The IO buffers on the iobuf list and the shadow buffers on this
         * list match each other one for one at all times. [j_list_lock]
         */

        struct journal_head     *t_shadow_list;

        /*
         * Doubly-linked circular list of control buffers being written to the
         * log. [j_list_lock]
         */

        struct journal_head     *t_log_list;

        /*
         * Protects info related to handles
         */

        spinlock_t              t_handle_lock;

        /*
         * Number of outstanding updates running on this transaction
         * [t_handle_lock]
         */

        int                     t_updates;

        /*
         * Number of buffers reserved for use by all handles in this transaction
         * handle but not yet modified. [t_handle_lock]
         */

        int                     t_outstanding_credits;

        /*
         * Forward and backward links for the circular list of all transactions
         * awaiting checkpoint. [j_list_lock]
         */

        transaction_t           *t_cpnext, *t_cpprev;

        /*
         * When will the transaction expire (become due for commit), in jiffies?
         * [no locking]
         */

        unsigned long           t_expires;

        /*
         * How many handles used this transaction? [t_handle_lock]
         */

        int t_handle_count;

        /*
         * Protects the callback list
         */

        spinlock_t              t_jcb_lock;
        /*
         * List of registered callback functions for this transaction.
         * Called when the transaction is committed. [t_jcb_lock]
         */

        struct list_head        t_jcb;
};

/**
 * struct journal_s - The journal_s type is the concrete type associated with
 *     journal_t.
 * @j_flags:  General journaling state flags
 * @j_errno:  Is there an outstanding uncleared error on the journal (from a
 *     prior abort)?
 * @j_sb_buffer: First part of superblock buffer
 * @j_superblock: Second part of superblock buffer
 * @j_format_version: Version of the superblock format
 * @j_barrier_count:  Number of processes waiting to create a barrier lock
 * @j_barrier: The barrier lock itself
 * @j_running_transaction: The current running transaction..
 * @j_committing_transaction: the transaction we are pushing to disk
 * @j_checkpoint_transactions: a linked circular list of all transactions
 *  waiting for checkpointing
 * @j_wait_transaction_locked: Wait queue for waiting for a locked transaction
 *  to start committing, or for a barrier lock to be released
 * @j_wait_logspace: Wait queue for waiting for checkpointing to complete
 * @j_wait_done_commit: Wait queue for waiting for commit to complete
 * @j_wait_checkpoint:  Wait queue to trigger checkpointing
 * @j_wait_commit: Wait queue to trigger commit
 * @j_wait_updates: Wait queue to wait for updates to complete
 * @j_checkpoint_sem: Semaphore for locking against concurrent checkpoints
 * @j_head: Journal head - identifies the first unused block in the journal
 * @j_tail: Journal tail - identifies the oldest still-used block in the
 *  journal.
 * @j_free: Journal free - how many free blocks are there in the journal?
 * @j_first: The block number of the first usable block
 * @j_last: The block number one beyond the last usable block
 * @j_dev: Device where we store the journal
 * @j_blocksize: blocksize for the location where we store the journal.
 * @j_blk_offset: starting block offset for into the device where we store the
 *     journal
 * @j_fs_dev: Device which holds the client fs.  For internal journal this will
 *     be equal to j_dev
 * @j_maxlen: Total maximum capacity of the journal region on disk.
 * @j_inode: Optional inode where we store the journal.  If present, all journal
 *     block numbers are mapped into this inode via bmap().
 * @j_tail_sequence:  Sequence number of the oldest transaction in the log
 * @j_transaction_sequence: Sequence number of the next transaction to grant
 * @j_commit_sequence: Sequence number of the most recently committed
 *  transaction
 * @j_commit_request: Sequence number of the most recent transaction wanting
 *     commit
 * @j_uuid: Uuid of client object.
 * @j_task: Pointer to the current commit thread for this journal
 * @j_max_transaction_buffers:  Maximum number of metadata buffers to allow in a
 *     single compound commit transaction
 * @j_commit_interval: What is the maximum transaction lifetime before we begin
 *  a commit?
 * @j_commit_timer:  The timer used to wakeup the commit thread
 * @j_revoke: The revoke table - maintains the list of revoked blocks in the
 *     current transaction.
 */


struct journal_s
{
        /* General journaling state flags [j_state_lock] */
        unsigned long           j_flags;

        /*
         * Is there an outstanding uncleared error on the journal (from a prior
         * abort)? [j_state_lock]
         */

        int                     j_errno;

        /* The superblock buffer */
        struct buffer_head      *j_sb_buffer;
        journal_superblock_t    *j_superblock;

        /* Version of the superblock format */
        int                     j_format_version;

        /*
         * Protect the various scalars in the journal
         */

        spinlock_t              j_state_lock;

        /*
         * Number of processes waiting to create a barrier lock [j_state_lock]
         */

        int                     j_barrier_count;

        /* The barrier lock itself */
        struct semaphore        j_barrier;

        /*
         * Transactions: The current running transaction...
         * [j_state_lock] [caller holding open handle]
         */

        transaction_t           *j_running_transaction;

        /*
         * the transaction we are pushing to disk
         * [j_state_lock] [caller holding open handle]
         */

        transaction_t           *j_committing_transaction;

        /*
         * ... and a linked circular list of all transactions waiting for
         * checkpointing. [j_list_lock]
         */

        transaction_t           *j_checkpoint_transactions;

        /*
         * Wait queue for waiting for a locked transaction to start committing,
         * or for a barrier lock to be released
         */

        wait_queue_head_t       j_wait_transaction_locked;

        /* Wait queue for waiting for checkpointing to complete */
        wait_queue_head_t       j_wait_logspace;

        /* Wait queue for waiting for commit to complete */
        wait_queue_head_t       j_wait_done_commit;

        /* Wait queue to trigger checkpointing */
        wait_queue_head_t       j_wait_checkpoint;

        /* Wait queue to trigger commit */
        wait_queue_head_t       j_wait_commit;

        /* Wait queue to wait for updates to complete */
        wait_queue_head_t       j_wait_updates;

        /* Semaphore for locking against concurrent checkpoints */
        struct semaphore        j_checkpoint_sem;

        /*
         * Journal head: identifies the first unused block in the journal.
         * [j_state_lock]
         */

        unsigned long           j_head;

        /*
         * Journal tail: identifies the oldest still-used block in the journal.
         * [j_state_lock]
         */

        unsigned long           j_tail;

        /*
         * Journal free: how many free blocks are there in the journal?
         * [j_state_lock]
         */

        unsigned long           j_free;

        /*
         * Journal start and end: the block numbers of the first usable block
         * and one beyond the last usable block in the journal. [j_state_lock]
         */

        unsigned long           j_first;
        unsigned long           j_last;

        /*
         * Device, blocksize and starting block offset for the location where we
         * store the journal.
         */

        struct block_device     *j_dev;
        int                     j_blocksize;
        unsigned int            j_blk_offset;

        /*
         * Device which holds the client fs.  For internal journal this will be
         * equal to j_dev.
         */

        struct block_device     *j_fs_dev;

        /* Total maximum capacity of the journal region on disk. */
        unsigned int            j_maxlen;

        /*
         * Protects the buffer lists and internal buffer state.
         */

        spinlock_t              j_list_lock;

        /* Optional inode where we store the journal.  If present, all */
        /* journal block numbers are mapped into this inode via */
        /* bmap(). */
        struct inode            *j_inode;

        /*
         * Sequence number of the oldest transaction in the log [j_state_lock]
         */

        tid_t                   j_tail_sequence;

        /*
         * Sequence number of the next transaction to grant [j_state_lock]
         */

        tid_t                   j_transaction_sequence;

        /*
         * Sequence number of the most recently committed transaction
         * [j_state_lock].
         */

        tid_t                   j_commit_sequence;

        /*
         * Sequence number of the most recent transaction wanting commit
         * [j_state_lock]
         */

        tid_t                   j_commit_request;

        /*
         * Journal uuid: identifies the object (filesystem, LVM volume etc)
         * backed by this journal.  This will eventually be replaced by an array
         * of uuids, allowing us to index multiple devices within a single
         * journal and to perform atomic updates across them.
         */

        __u8                    j_uuid[16];

        /* Pointer to the current commit thread for this journal */
        struct task_struct      *j_task;

        /*
         * Maximum number of metadata buffers to allow in a single compound
         * commit transaction
         */

        int                     j_max_transaction_buffers;

        /*
         * What is the maximum transaction lifetime before we begin a commit?
         */

        unsigned long           j_commit_interval;

        /* The timer used to wakeup the commit thread: */
        struct timer_list       *j_commit_timer;

        /*
         * The revoke table: maintains the list of revoked blocks in the
         * current transaction.  [j_revoke_lock]
         */

        spinlock_t              j_revoke_lock;
        struct jbd_revoke_table_s *j_revoke;
        struct jbd_revoke_table_s *j_revoke_table[2];

        /*
         * An opaque pointer to fs-private information.  ext3 puts its
         * superblock pointer here
         */

        void *j_private;
};

/*
 * Journal flag definitions
 */

#define JFS_UNMOUNT     0x001   /* Journal thread is being destroyed */
#define JFS_ABORT       0x002   /* Journaling has been aborted for errors. */
#define JFS_ACK_ERR     0x004   /* The errno in the sb has been acked */
#define JFS_FLUSHED     0x008   /* The journal superblock has been flushed */
#define JFS_LOADED      0x010   /* The journal superblock has been loaded */

/*
 * Function declarations for the journaling transaction and buffer
 * management
 */


/* Filing buffers */
extern void journal_unfile_buffer(journal_t *, struct journal_head *);
extern void __journal_unfile_buffer(struct journal_head *);
extern void __journal_refile_buffer(struct journal_head *);
extern void journal_refile_buffer(journal_t *, struct journal_head *);
extern void __journal_file_buffer(struct journal_head *, transaction_t *, int);
extern void __journal_free_buffer(struct journal_head *bh);
extern void journal_file_buffer(struct journal_head *, transaction_t *, int);
extern void __journal_clean_data_list(transaction_t *transaction);

/* Log buffer allocation */
extern struct journal_head * journal_get_descriptor_buffer(journal_t *);
int journal_next_log_block(journal_t *, unsigned long *);

/* Commit management */
extern void journal_commit_transaction(journal_t *);

/* Checkpoint list management */
int __journal_clean_checkpoint_list(journal_t *journal);
void __journal_remove_checkpoint(struct journal_head *);
void __journal_insert_checkpoint(struct journal_head *, transaction_t *);

/* Buffer IO */
extern int
journal_write_metadata_buffer(transaction_t       *transaction,
                              struct journal_head  *jh_in,
                              struct journal_head **jh_out,
                              int                  blocknr);

/* Transaction locking */
extern void             __wait_on_journal (journal_t *);

/*
 * Journal locking.
 *
 * We need to lock the journal during transaction state changes so that nobody
 * ever tries to take a handle on the running transaction while we are in the
 * middle of moving it to the commit phase.  j_state_lock does this.
 *
 * Note that the locking is completely interrupt unsafe.  We never touch
 * journal structures from interrupts.
 */


static inline handle_t *journal_current_handle(void)
{
        return current->journal_info;
}

/* The journaling code user interface:
 *
 * Create and destroy handles
 * Register buffer modifications against the current transaction.
 */


extern handle_t *journal_start(journal_t *, int nblocks);
extern int       journal_restart (handle_t *, int nblocks);
extern int       journal_extend (handle_t *, int nblocks);
extern int       journal_get_write_access(handle_t *, struct buffer_head *,
                                                int *credits);
extern int       journal_get_create_access (handle_t *, struct buffer_head *);
extern int       journal_get_undo_access(handle_t *, struct buffer_head *,
                                                int *credits);
extern int       journal_dirty_data (handle_t *, struct buffer_head *);
extern int       journal_dirty_metadata (handle_t *, struct buffer_head *);
extern void      journal_release_buffer (handle_t *, struct buffer_head *,
                                                int credits);
extern void      journal_forget (handle_t *, struct buffer_head *);
extern void      journal_sync_buffer (struct buffer_head *);
extern int       journal_invalidatepage(journal_t *,
                                struct page *, unsigned long);
extern int       journal_try_to_free_buffers(journal_t *, struct page *, int);
extern int       journal_stop(handle_t *);
extern int       journal_flush (journal_t *);
extern void      journal_callback_set(handle_t *handle,
                                      void (*fn)(struct journal_callback *,int),
                                      struct journal_callback *jcb);

extern void      journal_lock_updates (journal_t *);
extern void      journal_unlock_updates (journal_t *);

extern journal_t * journal_init_dev(struct block_device *bdev,
                                struct block_device *fs_dev,
                                int start, int len, int bsize);
extern journal_t * journal_init_inode (struct inode *);
extern int         journal_update_format (journal_t *);
extern int         journal_check_used_features
                   (journal_t *, unsigned long, unsigned long, unsigned long);
extern int         journal_check_available_features
                   (journal_t *, unsigned long, unsigned long, unsigned long);
extern int         journal_set_features
                   (journal_t *, unsigned long, unsigned long, unsigned long);
extern int         journal_create     (journal_t *);
extern int         journal_load       (journal_t *journal);
extern void        journal_destroy    (journal_t *);
extern int         journal_recover    (journal_t *journal);
extern int         journal_wipe       (journal_t *, int);
extern int         journal_skip_recovery        (journal_t *);
extern void        journal_update_superblock    (journal_t *, int);
extern void        __journal_abort_hard (journal_t *);
extern void        __journal_abort_soft (journal_t *, int);
extern void        journal_abort      (journal_t *, int);
extern int         journal_errno      (journal_t *);
extern void        journal_ack_err    (journal_t *);
extern int         journal_clear_err  (journal_t *);
extern int         journal_bmap(journal_t *, unsigned long, unsigned long *);
extern int         journal_force_commit(journal_t *);

/*
 * journal_head management
 */

struct journal_head *journal_add_journal_head(struct buffer_head *bh);
struct journal_head *journal_grab_journal_head(struct buffer_head *bh);
void journal_remove_journal_head(struct buffer_head *bh);
void journal_put_journal_head(struct journal_head *jh);

/*
 * handle management
 */

extern kmem_cache_t *jbd_handle_cache;

static inline handle_t *jbd_alloc_handle(int gfp_flags)
{
        return kmem_cache_alloc(jbd_handle_cache, gfp_flags);
}

static inline void jbd_free_handle(handle_t *handle)
{
        kmem_cache_free(jbd_handle_cache, handle);
}

/* Primary revoke support */
#define JOURNAL_REVOKE_DEFAULT_HASH 256
extern int         journal_init_revoke(journal_t *, int);
extern void        journal_destroy_revoke_caches(void);
extern int         journal_init_revoke_caches(void);

extern void        journal_destroy_revoke(journal_t *);
extern int         journal_revoke (handle_t *,
                                unsigned long, struct buffer_head *);
extern int         journal_cancel_revoke(handle_t *, struct journal_head *);
extern void        journal_write_revoke_records(journal_t *, transaction_t *);

/* Recovery revoke support */
extern int      journal_set_revoke(journal_t *, unsigned long, tid_t);
extern int      journal_test_revoke(journal_t *, unsigned long, tid_t);
extern void     journal_clear_revoke(journal_t *);
extern void     journal_brelse_array(struct buffer_head *b[], int n);
extern void     journal_switch_revoke_table(journal_t *journal);

/*
 * The log thread user interface:
 *
 * Request space in the current transaction, and force transaction commit
 * transitions on demand.
 */


int __log_space_left(journal_t *); /* Called with journal locked */
int log_start_commit(journal_t *journal, tid_t tid);
int __log_start_commit(journal_t *journal, tid_t tid);
int journal_start_commit(journal_t *journal, tid_t *tid);
int log_wait_commit(journal_t *journal, tid_t tid);
int log_do_checkpoint(journal_t *journal);

void __log_wait_for_space(journal_t *journal);
extern void     __journal_drop_transaction(journal_t *, transaction_t *);
extern int      cleanup_journal_tail(journal_t *);

/* Debugging code only: */

#define jbd_ENOSYS() \
do {                                                                  \
        printk (KERN_ERR "JBD unimplemented function " __FUNCTION__); \
        current->state = TASK_UNINTERRUPTIBLE;                        \
        schedule();                                                   \
} while (1)


/*
 * is_journal_abort
 *
 * Simple test wrapper function to test the JFS_ABORT state flag.  This
 * bit, when set, indicates that we have had a fatal error somewhere,
 * either inside the journaling layer or indicated to us by the client
 * (eg. ext3), and that we and should not commit any further
 * transactions.  
 */


static inline int is_journal_aborted(journal_t *journal)
{
        return journal->j_flags & JFS_ABORT;
}

static inline int is_handle_aborted(handle_t *handle)
{
        if (handle->h_aborted)
                return 1;
        return is_journal_aborted(handle->h_transaction->t_journal);
}

static inline void journal_abort_handle(handle_t *handle)
{
        handle->h_aborted = 1;
}

#endif /* __KERNEL__   */

/* Comparison functions for transaction IDs: perform comparisons using
 * modulo arithmetic so that they work over sequence number wraps. */


static inline int tid_gt(tid_t x, tid_t y)
{
        int difference = (x - y);
        return (difference > 0);
}

static inline int tid_geq(tid_t x, tid_t y)
{
        int difference = (x - y);
        return (difference >= 0);
}

extern int journal_blocks_per_page(struct inode *inode);

/*
 * Return the minimum number of blocks which must be free in the journal
 * before a new transaction may be started.  Must be called under j_state_lock.
 */

static inline int jbd_space_needed(journal_t *journal)
{
        int nblocks = journal->j_max_transaction_buffers;
        if (journal->j_committing_transaction)
                nblocks += journal->j_committing_transaction->
                                        t_outstanding_credits;
        return nblocks;
}

/*
 * Definitions which augment the buffer_head layer
 */


/* journaling buffer types */
#define BJ_None         0       /* Not journaled */
#define BJ_SyncData     1       /* Normal data: flush before commit */
#define BJ_Metadata     2       /* Normal journaled metadata */
#define BJ_Forget       3       /* Buffer superseded by this transaction */
#define BJ_IO           4       /* Buffer is for temporary IO use */
#define BJ_Shadow       5       /* Buffer contents being shadowed to the log */
#define BJ_LogCtl       6       /* Buffer contains log descriptors */
#define BJ_Reserved     7       /* Buffer is reserved for access by journal */
#define BJ_Types        8
 
extern int jbd_blocks_per_page(struct inode *inode);

#ifdef __KERNEL__

#ifdef CONFIG_SMP
#define assert_spin_locked(lock)        J_ASSERT(spin_is_locked(lock))
#else
#define assert_spin_locked(lock)        do {} while(0)
#endif

#define buffer_trace_init(bh)   do {} while (0)
#define print_buffer_fields(bh) do {} while (0)
#define print_buffer_trace(bh)  do {} while (0)
#define BUFFER_TRACE(bh, info)  do {} while (0)
#define BUFFER_TRACE2(bh, bh2, info)    do {} while (0)
#define JBUFFER_TRACE(jh, info) do {} while (0)

#endif  /* __KERNEL__ */

#endif  /* CONFIG_JBD || CONFIG_JBD_MODULE || !__KERNEL__ */

/*
 * Compatibility no-ops which allow the kernel to compile without CONFIG_JBD
 * go here.
 */


#if defined(__KERNEL__) && !(defined(CONFIG_JBD) || defined(CONFIG_JBD_MODULE))

#define J_ASSERT(expr)                  do {} while (0)
#define J_ASSERT_BH(bh, expr)           do {} while (0)
#define buffer_jbd(bh)                  0
#define journal_buffer_journal_lru(bh)  0

#endif  /* defined(__KERNEL__) && !defined(CONFIG_JBD) */
#endif  /* _LINUX_JBD_H */