/*
 * Copyright (c) 2000-2006 Silicon Graphics, Inc.
 * All Rights Reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it would be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_types.h"
#include "xfs_bit.h"
#include "xfs_log.h"
#include "xfs_inum.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_mount.h"
#include "xfs_error.h"
#include "xfs_bmap_btree.h"
#include "xfs_alloc_btree.h"
#include "xfs_ialloc_btree.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_inode_item.h"
#include "xfs_alloc.h"
#include "xfs_ialloc.h"
#include "xfs_log_priv.h"
#include "xfs_buf_item.h"
#include "xfs_log_recover.h"
#include "xfs_extfree_item.h"
#include "xfs_trans_priv.h"
#include "xfs_quota.h"
#include "xfs_utils.h"
#include "xfs_trace.h"

STATIC int
xlog_find_zeroed(
        struct xlog     *,
        xfs_daddr_t     *);
STATIC int
xlog_clear_stale_blocks(
        struct xlog     *,
        xfs_lsn_t);
#if defined(DEBUG)
STATIC void
xlog_recover_check_summary(
        struct xlog *);
#else
#define xlog_recover_check_summary(log)
#endif

/*
 * This structure is used during recovery to record the buf log items which
 * have been canceled and should not be replayed.
 */
struct xfs_buf_cancel {
        xfs_daddr_t             bc_blkno;
        uint                    bc_len;
        int                     bc_refcount;
        struct list_head        bc_list;
};

/*
 * Sector aligned buffer routines for buffer create/read/write/access
 */

/*
 * Verify the given count of basic blocks is valid number of blocks
 * to specify for an operation involving the given XFS log buffer.
 * Returns nonzero if the count is valid, 0 otherwise.
 */

static inline int
xlog_buf_bbcount_valid(
        struct xlog     *log,
        int             bbcount)
{
        return bbcount > 0 && bbcount <= log->l_logBBsize;
}

/*
 * Allocate a buffer to hold log data.  The buffer needs to be able
 * to map to a range of nbblks basic blocks at any valid (basic
 * block) offset within the log.
 */
STATIC xfs_buf_t *
xlog_get_bp(
        struct xlog     *log,
        int             nbblks)
{
        struct xfs_buf  *bp;

        if (!xlog_buf_bbcount_valid(log, nbblks)) {
                xfs_warn(log->l_mp, "Invalid block length (0x%x) for buffer",
                        nbblks);
                XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_HIGH, log->l_mp);
                return NULL;
        }

        /*
         * We do log I/O in units of log sectors (a power-of-2
         * multiple of the basic block size), so we round up the
         * requested size to accommodate the basic blocks required
         * for complete log sectors.
         *
         * In addition, the buffer may be used for a non-sector-
         * aligned block offset, in which case an I/O of the
         * requested size could extend beyond the end of the
         * buffer.  If the requested size is only 1 basic block it
         * will never straddle a sector boundary, so this won't be
         * an issue.  Nor will this be a problem if the log I/O is
         * done in basic blocks (sector size 1).  But otherwise we
         * extend the buffer by one extra log sector to ensure
         * there's space to accommodate this possibility.
         */
        if (nbblks > 1 && log->l_sectBBsize > 1)
                nbblks += log->l_sectBBsize;
        nbblks = round_up(nbblks, log->l_sectBBsize);

        bp = xfs_buf_get_uncached(log->l_mp->m_logdev_targp, nbblks, 0);
        if (bp)
                xfs_buf_unlock(bp);
        return bp;
}

STATIC void
xlog_put_bp(
        xfs_buf_t       *bp)
{
        xfs_buf_free(bp);
}

/*
 * Return the address of the start of the given block number's data
 * in a log buffer.  The buffer covers a log sector-aligned region.
 */
STATIC xfs_caddr_t
xlog_align(
        struct xlog     *log,
        xfs_daddr_t     blk_no,
        int             nbblks,
        struct xfs_buf  *bp)
{
        xfs_daddr_t     offset = blk_no & ((xfs_daddr_t)log->l_sectBBsize - 1);

        ASSERT(offset + nbblks <= bp->b_length);
        return bp->b_addr + BBTOB(offset);
}


/*
 * nbblks should be uint, but oh well.  Just want to catch that 32-bit length.
 */
STATIC int
xlog_bread_noalign(
        struct xlog     *log,
        xfs_daddr_t     blk_no,
        int             nbblks,
        struct xfs_buf  *bp)
{
        int             error;

        if (!xlog_buf_bbcount_valid(log, nbblks)) {
                xfs_warn(log->l_mp, "Invalid block length (0x%x) for buffer",
                        nbblks);
                XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_HIGH, log->l_mp);
                return EFSCORRUPTED;
        }

        blk_no = round_down(blk_no, log->l_sectBBsize);
        nbblks = round_up(nbblks, log->l_sectBBsize);

        ASSERT(nbblks > 0);
        ASSERT(nbblks <= bp->b_length);

        XFS_BUF_SET_ADDR(bp, log->l_logBBstart + blk_no);
        XFS_BUF_READ(bp);
        bp->b_io_length = nbblks;
        bp->b_error = 0;

        xfsbdstrat(log->l_mp, bp);
        error = xfs_buf_iowait(bp);
        if (error)
                xfs_buf_ioerror_alert(bp, __func__);
        return error;
}

STATIC int
xlog_bread(
        struct xlog     *log,
        xfs_daddr_t     blk_no,
        int             nbblks,
        struct xfs_buf  *bp,
        xfs_caddr_t     *offset)
{
        int             error;

        error = xlog_bread_noalign(log, blk_no, nbblks, bp);
        if (error)
                return error;

        *offset = xlog_align(log, blk_no, nbblks, bp);
        return 0;
}

/*
 * Read at an offset into the buffer. Returns with the buffer in it's original
 * state regardless of the result of the read.
 */
STATIC int
xlog_bread_offset(
        struct xlog     *log,
        xfs_daddr_t     blk_no,         /* block to read from */
        int             nbblks,         /* blocks to read */
        struct xfs_buf  *bp,
        xfs_caddr_t     offset)
{
        xfs_caddr_t     orig_offset = bp->b_addr;
        int             orig_len = BBTOB(bp->b_length);
        int             error, error2;

        error = xfs_buf_associate_memory(bp, offset, BBTOB(nbblks));
        if (error)
                return error;

        error = xlog_bread_noalign(log, blk_no, nbblks, bp);

        /* must reset buffer pointer even on error */
        error2 = xfs_buf_associate_memory(bp, orig_offset, orig_len);
        if (error)
                return error;
        return error2;
}

/*
 * Write out the buffer at the given block for the given number of blocks.
 * The buffer is kept locked across the write and is returned locked.
 * This can only be used for synchronous log writes.
 */
STATIC int
xlog_bwrite(
        struct xlog     *log,
        xfs_daddr_t     blk_no,
        int             nbblks,
        struct xfs_buf  *bp)
{
        int             error;

        if (!xlog_buf_bbcount_valid(log, nbblks)) {
                xfs_warn(log->l_mp, "Invalid block length (0x%x) for buffer",
                        nbblks);
                XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_HIGH, log->l_mp);
                return EFSCORRUPTED;
        }

        blk_no = round_down(blk_no, log->l_sectBBsize);
        nbblks = round_up(nbblks, log->l_sectBBsize);

        ASSERT(nbblks > 0);
        ASSERT(nbblks <= bp->b_length);

        XFS_BUF_SET_ADDR(bp, log->l_logBBstart + blk_no);
        XFS_BUF_ZEROFLAGS(bp);
        xfs_buf_hold(bp);
        xfs_buf_lock(bp);
        bp->b_io_length = nbblks;
        bp->b_error = 0;

        error = xfs_bwrite(bp);
        if (error)
                xfs_buf_ioerror_alert(bp, __func__);
        xfs_buf_relse(bp);
        return error;
}

#ifdef DEBUG
/*
 * dump debug superblock and log record information
 */
STATIC void
xlog_header_check_dump(
        xfs_mount_t             *mp,
        xlog_rec_header_t       *head)
{
        xfs_debug(mp, "%s:  SB : uuid = %pU, fmt = %d\n",
                __func__, &mp->m_sb.sb_uuid, XLOG_FMT);
        xfs_debug(mp, "    log : uuid = %pU, fmt = %d\n",
                &head->h_fs_uuid, be32_to_cpu(head->h_fmt));
}
#else
#define xlog_header_check_dump(mp, head)
#endif

/*
 * check log record header for recovery
 */
STATIC int
xlog_header_check_recover(
        xfs_mount_t             *mp,
        xlog_rec_header_t       *head)
{
        ASSERT(head->h_magicno == cpu_to_be32(XLOG_HEADER_MAGIC_NUM));

        /*
         * IRIX doesn't write the h_fmt field and leaves it zeroed
         * (XLOG_FMT_UNKNOWN). This stops us from trying to recover
         * a dirty log created in IRIX.
         */
        if (unlikely(head->h_fmt != cpu_to_be32(XLOG_FMT))) {
                xfs_warn(mp,
        "dirty log written in incompatible format - can't recover");
                xlog_header_check_dump(mp, head);
                XFS_ERROR_REPORT("xlog_header_check_recover(1)",
                                 XFS_ERRLEVEL_HIGH, mp);
                return XFS_ERROR(EFSCORRUPTED);
        } else if (unlikely(!uuid_equal(&mp->m_sb.sb_uuid, &head->h_fs_uuid))) {
                xfs_warn(mp,
        "dirty log entry has mismatched uuid - can't recover");
                xlog_header_check_dump(mp, head);
                XFS_ERROR_REPORT("xlog_header_check_recover(2)",
                                 XFS_ERRLEVEL_HIGH, mp);
                return XFS_ERROR(EFSCORRUPTED);
        }
        return 0;
}

/*
 * read the head block of the log and check the header
 */
STATIC int
xlog_header_check_mount(
        xfs_mount_t             *mp,
        xlog_rec_header_t       *head)
{
        ASSERT(head->h_magicno == cpu_to_be32(XLOG_HEADER_MAGIC_NUM));

        if (uuid_is_nil(&head->h_fs_uuid)) {
                /*
                 * IRIX doesn't write the h_fs_uuid or h_fmt fields. If
                 * h_fs_uuid is nil, we assume this log was last mounted
                 * by IRIX and continue.
                 */
                xfs_warn(mp, "nil uuid in log - IRIX style log");
        } else if (unlikely(!uuid_equal(&mp->m_sb.sb_uuid, &head->h_fs_uuid))) {
                xfs_warn(mp, "log has mismatched uuid - can't recover");
                xlog_header_check_dump(mp, head);
                XFS_ERROR_REPORT("xlog_header_check_mount",
                                 XFS_ERRLEVEL_HIGH, mp);
                return XFS_ERROR(EFSCORRUPTED);
        }
        return 0;
}

STATIC void
xlog_recover_iodone(
        struct xfs_buf  *bp)
{
        if (bp->b_error) {
                /*
                 * We're not going to bother about retrying
                 * this during recovery. One strike!
                 */
                xfs_buf_ioerror_alert(bp, __func__);
                xfs_force_shutdown(bp->b_target->bt_mount,
                                        SHUTDOWN_META_IO_ERROR);
        }
        bp->b_iodone = NULL;
        xfs_buf_ioend(bp, 0);
}

/*
 * This routine finds (to an approximation) the first block in the physical
 * log which contains the given cycle.  It uses a binary search algorithm.
 * Note that the algorithm can not be perfect because the disk will not
 * necessarily be perfect.
 */
STATIC int
xlog_find_cycle_start(
        struct xlog     *log,
        struct xfs_buf  *bp,
        xfs_daddr_t     first_blk,
        xfs_daddr_t     *last_blk,
        uint            cycle)
{
        xfs_caddr_t     offset;
        xfs_daddr_t     mid_blk;
        xfs_daddr_t     end_blk;
        uint            mid_cycle;
        int             error;

        end_blk = *last_blk;
        mid_blk = BLK_AVG(first_blk, end_blk);
        while (mid_blk != first_blk && mid_blk != end_blk) {
                error = xlog_bread(log, mid_blk, 1, bp, &offset);
                if (error)
                        return error;
                mid_cycle = xlog_get_cycle(offset);
                if (mid_cycle == cycle)
                        end_blk = mid_blk;   /* last_half_cycle == mid_cycle */
                else
                        first_blk = mid_blk; /* first_half_cycle == mid_cycle */
                mid_blk = BLK_AVG(first_blk, end_blk);
        }
        ASSERT((mid_blk == first_blk && mid_blk+1 == end_blk) ||
               (mid_blk == end_blk && mid_blk-1 == first_blk));

        *last_blk = end_blk;

        return 0;
}

/*
 * Check that a range of blocks does not contain stop_on_cycle_no.
 * Fill in *new_blk with the block offset where such a block is
 * found, or with -1 (an invalid block number) if there is no such
 * block in the range.  The scan needs to occur from front to back
 * and the pointer into the region must be updated since a later
 * routine will need to perform another test.
 */
STATIC int
xlog_find_verify_cycle(
        struct xlog     *log,
        xfs_daddr_t     start_blk,
        int             nbblks,
        uint            stop_on_cycle_no,
        xfs_daddr_t     *new_blk)
{
        xfs_daddr_t     i, j;
        uint            cycle;
        xfs_buf_t       *bp;
        xfs_daddr_t     bufblks;
        xfs_caddr_t     buf = NULL;
        int             error = 0;

        /*
         * Greedily allocate a buffer big enough to handle the full
         * range of basic blocks we'll be examining.  If that fails,
         * try a smaller size.  We need to be able to read at least
         * a log sector, or we're out of luck.
         */
        bufblks = 1 << ffs(nbblks);
        while (bufblks > log->l_logBBsize)
                bufblks >>= 1;
        while (!(bp = xlog_get_bp(log, bufblks))) {
                bufblks >>= 1;
                if (bufblks < log->l_sectBBsize)
                        return ENOMEM;
        }

        for (i = start_blk; i < start_blk + nbblks; i += bufblks) {
                int     bcount;

                bcount = min(bufblks, (start_blk + nbblks - i));

                error = xlog_bread(log, i, bcount, bp, &buf);
                if (error)
                        goto out;

                for (j = 0; j < bcount; j++) {
                        cycle = xlog_get_cycle(buf);
                        if (cycle == stop_on_cycle_no) {
                                *new_blk = i+j;
                                goto out;
                        }

                        buf += BBSIZE;
                }
        }

        *new_blk = -1;

out:
        xlog_put_bp(bp);
        return error;
}

/*
 * Potentially backup over partial log record write.
 *
 * In the typical case, last_blk is the number of the block directly after
 * a good log record.  Therefore, we subtract one to get the block number
 * of the last block in the given buffer.  extra_bblks contains the number
 * of blocks we would have read on a previous read.  This happens when the
 * last log record is split over the end of the physical log.
 *
 * extra_bblks is the number of blocks potentially verified on a previous
 * call to this routine.
 */
STATIC int
xlog_find_verify_log_record(
        struct xlog             *log,
        xfs_daddr_t             start_blk,
        xfs_daddr_t             *last_blk,
        int                     extra_bblks)
{
        xfs_daddr_t             i;
        xfs_buf_t               *bp;
        xfs_caddr_t             offset = NULL;
        xlog_rec_header_t       *head = NULL;
        int                     error = 0;
        int                     smallmem = 0;
        int                     num_blks = *last_blk - start_blk;
        int                     xhdrs;

        ASSERT(start_blk != 0 || *last_blk != start_blk);

        if (!(bp = xlog_get_bp(log, num_blks))) {
                if (!(bp = xlog_get_bp(log, 1)))
                        return ENOMEM;
                smallmem = 1;
        } else {
                error = xlog_bread(log, start_blk, num_blks, bp, &offset);
                if (error)
                        goto out;
                offset += ((num_blks - 1) << BBSHIFT);
        }

        for (i = (*last_blk) - 1; i >= 0; i--) {
                if (i < start_blk) {
                        /* valid log record not found */
                        xfs_warn(log->l_mp,
                "Log inconsistent (didn't find previous header)");
                        ASSERT(0);
                        error = XFS_ERROR(EIO);
                        goto out;
                }

                if (smallmem) {
                        error = xlog_bread(log, i, 1, bp, &offset);
                        if (error)
                                goto out;
                }

                head = (xlog_rec_header_t *)offset;

                if (head->h_magicno == cpu_to_be32(XLOG_HEADER_MAGIC_NUM))
                        break;

                if (!smallmem)
                        offset -= BBSIZE;
        }

        /*
         * We hit the beginning of the physical log & still no header.  Return
         * to caller.  If caller can handle a return of -1, then this routine
         * will be called again for the end of the physical log.
         */
        if (i == -1) {
                error = -1;
                goto out;
        }

        /*
         * We have the final block of the good log (the first block
         * of the log record _before_ the head. So we check the uuid.
         */
        if ((error = xlog_header_check_mount(log->l_mp, head)))
                goto out;

        /*
         * We may have found a log record header before we expected one.
         * last_blk will be the 1st block # with a given cycle #.  We may end
         * up reading an entire log record.  In this case, we don't want to
         * reset last_blk.  Only when last_blk points in the middle of a log
         * record do we update last_blk.
         */
        if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
                uint    h_size = be32_to_cpu(head->h_size);

                xhdrs = h_size / XLOG_HEADER_CYCLE_SIZE;
                if (h_size % XLOG_HEADER_CYCLE_SIZE)
                        xhdrs++;
        } else {
                xhdrs = 1;
        }

        if (*last_blk - i + extra_bblks !=
            BTOBB(be32_to_cpu(head->h_len)) + xhdrs)
                *last_blk = i;

out:
        xlog_put_bp(bp);
        return error;
}

/*
 * Head is defined to be the point of the log where the next log write
 * write could go.  This means that incomplete LR writes at the end are
 * eliminated when calculating the head.  We aren't guaranteed that previous
 * LR have complete transactions.  We only know that a cycle number of
 * current cycle number -1 won't be present in the log if we start writing
 * from our current block number.
 *
 * last_blk contains the block number of the first block with a given
 * cycle number.
 *
 * Return: zero if normal, non-zero if error.
 */
STATIC int
xlog_find_head(
        struct xlog     *log,
        xfs_daddr_t     *return_head_blk)
{
        xfs_buf_t       *bp;
        xfs_caddr_t     offset;
        xfs_daddr_t     new_blk, first_blk, start_blk, last_blk, head_blk;
        int             num_scan_bblks;
        uint            first_half_cycle, last_half_cycle;
        uint            stop_on_cycle;
        int             error, log_bbnum = log->l_logBBsize;

        /* Is the end of the log device zeroed? */
        if ((error = xlog_find_zeroed(log, &first_blk)) == -1) {
                *return_head_blk = first_blk;

                /* Is the whole lot zeroed? */
                if (!first_blk) {
                        /* Linux XFS shouldn't generate totally zeroed logs -
                         * mkfs etc write a dummy unmount record to a fresh
                         * log so we can store the uuid in there
                         */
                        xfs_warn(log->l_mp, "totally zeroed log");
                }

                return 0;
        } else if (error) {
                xfs_warn(log->l_mp, "empty log check failed");
                return error;
        }

        first_blk = 0;                  /* get cycle # of 1st block */
        bp = xlog_get_bp(log, 1);
        if (!bp)
                return ENOMEM;

        error = xlog_bread(log, 0, 1, bp, &offset);
        if (error)
                goto bp_err;

        first_half_cycle = xlog_get_cycle(offset);

        last_blk = head_blk = log_bbnum - 1;    /* get cycle # of last block */
        error = xlog_bread(log, last_blk, 1, bp, &offset);
        if (error)
                goto bp_err;

        last_half_cycle = xlog_get_cycle(offset);
        ASSERT(last_half_cycle != 0);

        /*
         * If the 1st half cycle number is equal to the last half cycle number,
         * then the entire log is stamped with the same cycle number.  In this
         * case, head_blk can't be set to zero (which makes sense).  The below
         * math doesn't work out properly with head_blk equal to zero.  Instead,
         * we set it to log_bbnum which is an invalid block number, but this
         * value makes the math correct.  If head_blk doesn't changed through
         * all the tests below, *head_blk is set to zero at the very end rather
         * than log_bbnum.  In a sense, log_bbnum and zero are the same block
         * in a circular file.
         */
        if (first_half_cycle == last_half_cycle) {
                /*
                 * In this case we believe that the entire log should have
                 * cycle number last_half_cycle.  We need to scan backwards
                 * from the end verifying that there are no holes still
                 * containing last_half_cycle - 1.  If we find such a hole,
                 * then the start of that hole will be the new head.  The
                 * simple case looks like
                 *        x | x ... | x - 1 | x
                 * Another case that fits this picture would be
                 *        x | x + 1 | x ... | x
                 * In this case the head really is somewhere at the end of the
                 * log, as one of the latest writes at the beginning was
                 * incomplete.
                 * One more case is
                 *        x | x + 1 | x ... | x - 1 | x
                 * This is really the combination of the above two cases, and
                 * the head has to end up at the start of the x-1 hole at the
                 * end of the log.
                 *
                 * In the 256k log case, we will read from the beginning to the
                 * end of the log and search for cycle numbers equal to x-1.
                 * We don't worry about the x+1 blocks that we encounter,
                 * because we know that they cannot be the head since the log
                 * started with x.
                 */
                head_blk = log_bbnum;
                stop_on_cycle = last_half_cycle - 1;
        } else {
                /*
                 * In this case we want to find the first block with cycle
                 * number matching last_half_cycle.  We expect the log to be
                 * some variation on
                 *        x + 1 ... | x ... | x
                 * The first block with cycle number x (last_half_cycle) will
                 * be where the new head belongs.  First we do a binary search
                 * for the first occurrence of last_half_cycle.  The binary
                 * search may not be totally accurate, so then we scan back
                 * from there looking for occurrences of last_half_cycle before
                 * us.  If that backwards scan wraps around the beginning of
                 * the log, then we look for occurrences of last_half_cycle - 1
                 * at the end of the log.  The cases we're looking for look
                 * like
                 *                               v binary search stopped here
                 *        x + 1 ... | x | x + 1 | x ... | x
                 *                   ^ but we want to locate this spot
                 * or
                 *        <---------> less than scan distance
                 *        x + 1 ... | x ... | x - 1 | x
                 *                           ^ we want to locate this spot
                 */
                stop_on_cycle = last_half_cycle;
                if ((error = xlog_find_cycle_start(log, bp, first_blk,
                                                &head_blk, last_half_cycle)))
                        goto bp_err;
        }

        /*
         * Now validate the answer.  Scan back some number of maximum possible
         * blocks and make sure each one has the expected cycle number.  The
         * maximum is determined by the total possible amount of buffering
         * in the in-core log.  The following number can be made tighter if
         * we actually look at the block size of the filesystem.
         */
        num_scan_bblks = XLOG_TOTAL_REC_SHIFT(log);
        if (head_blk >= num_scan_bblks) {
                /*
                 * We are guaranteed that the entire check can be performed
                 * in one buffer.
                 */
                start_blk = head_blk - num_scan_bblks;
                if ((error = xlog_find_verify_cycle(log,
                                                start_blk, num_scan_bblks,
                                                stop_on_cycle, &new_blk)))
                        goto bp_err;
                if (new_blk != -1)
                        head_blk = new_blk;
        } else {                /* need to read 2 parts of log */
                /*
                 * We are going to scan backwards in the log in two parts.
                 * First we scan the physical end of the log.  In this part
                 * of the log, we are looking for blocks with cycle number
                 * last_half_cycle - 1.
                 * If we find one, then we know that the log starts there, as
                 * we've found a hole that didn't get written in going around
                 * the end of the physical log.  The simple case for this is
                 *        x + 1 ... | x ... | x - 1 | x
                 *        <---------> less than scan distance
                 * If all of the blocks at the end of the log have cycle number
                 * last_half_cycle, then we check the blocks at the start of
                 * the log looking for occurrences of last_half_cycle.  If we
                 * find one, then our current estimate for the location of the
                 * first occurrence of last_half_cycle is wrong and we move
                 * back to the hole we've found.  This case looks like
                 *        x + 1 ... | x | x + 1 | x ...
                 *                               ^ binary search stopped here
                 * Another case we need to handle that only occurs in 256k
                 * logs is
                 *        x + 1 ... | x ... | x+1 | x ...
                 *                   ^ binary search stops here
                 * In a 256k log, the scan at the end of the log will see the
                 * x + 1 blocks.  We need to skip past those since that is
                 * certainly not the head of the log.  By searching for
                 * last_half_cycle-1 we accomplish that.
                 */
                ASSERT(head_blk <= INT_MAX &&
                        (xfs_daddr_t) num_scan_bblks >= head_blk);
                start_blk = log_bbnum - (num_scan_bblks - head_blk);
                if ((error = xlog_find_verify_cycle(log, start_blk,
                                        num_scan_bblks - (int)head_blk,
                                        (stop_on_cycle - 1), &new_blk)))
                        goto bp_err;
                if (new_blk != -1) {
                        head_blk = new_blk;
                        goto validate_head;
                }

                /*
                 * Scan beginning of log now.  The last part of the physical
                 * log is good.  This scan needs to verify that it doesn't find
                 * the last_half_cycle.
                 */
                start_blk = 0;
                ASSERT(head_blk <= INT_MAX);
                if ((error = xlog_find_verify_cycle(log,
                                        start_blk, (int)head_blk,
                                        stop_on_cycle, &new_blk)))
                        goto bp_err;
                if (new_blk != -1)
                        head_blk = new_blk;
        }

validate_head:
        /*
         * Now we need to make sure head_blk is not pointing to a block in
         * the middle of a log record.
         */
        num_scan_bblks = XLOG_REC_SHIFT(log);
        if (head_blk >= num_scan_bblks) {
                start_blk = head_blk - num_scan_bblks; /* don't read head_blk */

                /* start ptr at last block ptr before head_blk */
                if ((error = xlog_find_verify_log_record(log, start_blk,
                                                        &head_blk, 0)) == -1) {
                        error = XFS_ERROR(EIO);
                        goto bp_err;
                } else if (error)
                        goto bp_err;
        } else {
                start_blk = 0;
                ASSERT(head_blk <= INT_MAX);
                if ((error = xlog_find_verify_log_record(log, start_blk,
                                                        &head_blk, 0)) == -1) {
                        /* We hit the beginning of the log during our search */
                        start_blk = log_bbnum - (num_scan_bblks - head_blk);
                        new_blk = log_bbnum;
                        ASSERT(start_blk <= INT_MAX &&
                                (xfs_daddr_t) log_bbnum-start_blk >= 0);
                        ASSERT(head_blk <= INT_MAX);
                        if ((error = xlog_find_verify_log_record(log,
                                                        start_blk, &new_blk,
                                                        (int)head_blk)) == -1) {
                                error = XFS_ERROR(EIO);
                                goto bp_err;
                        } else if (error)
                                goto bp_err;
                        if (new_blk != log_bbnum)
                                head_blk = new_blk;
                } else if (error)
                        goto bp_err;
        }

        xlog_put_bp(bp);
        if (head_blk == log_bbnum)
                *return_head_blk = 0;
        else
                *return_head_blk = head_blk;
        /*
         * When returning here, we have a good block number.  Bad block
         * means that during a previous crash, we didn't have a clean break
         * from cycle number N to cycle number N-1.  In this case, we need
         * to find the first block with cycle number N-1.
         */
        return 0;

 bp_err:
        xlog_put_bp(bp);

        if (error)
                xfs_warn(log->l_mp, "failed to find log head");
        return error;
}

/*
 * Find the sync block number or the tail of the log.
 *
 * This will be the block number of the last record to have its
 * associated buffers synced to disk.  Every log record header has
 * a sync lsn embedded in it.  LSNs hold block numbers, so it is easy
 * to get a sync block number.  The only concern is to figure out which
 * log record header to believe.
 *
 * The following algorithm uses the log record header with the largest
 * lsn.  The entire log record does not need to be valid.  We only care
 * that the header is valid.
 *
 * We could speed up search by using current head_blk buffer, but it is not
 * available.
 */
STATIC int
xlog_find_tail(
        struct xlog             *log,
        xfs_daddr_t             *head_blk,
        xfs_daddr_t             *tail_blk)
{
        xlog_rec_header_t       *rhead;
        xlog_op_header_t        *op_head;
        xfs_caddr_t             offset = NULL;
        xfs_buf_t               *bp;
        int                     error, i, found;
        xfs_daddr_t             umount_data_blk;
        xfs_daddr_t             after_umount_blk;
        xfs_lsn_t               tail_lsn;
        int                     hblks;

        found = 0;

        /*
         * Find previous log record
         */
        if ((error = xlog_find_head(log, head_blk)))
                return error;

        bp = xlog_get_bp(log, 1);
        if (!bp)
                return ENOMEM;
        if (*head_blk == 0) {                           /* special case */
                error = xlog_bread(log, 0, 1, bp, &offset);
                if (error)
                        goto done;

                if (xlog_get_cycle(offset) == 0) {
                        *tail_blk = 0;
                        /* leave all other log inited values alone */
                        goto done;
                }
        }

        /*
         * Search backwards looking for log record header block
         */
        ASSERT(*head_blk < INT_MAX);
        for (i = (int)(*head_blk) - 1; i >= 0; i--) {
                error = xlog_bread(log, i, 1, bp, &offset);
                if (error)
                        goto done;

                if (*(__be32 *)offset == cpu_to_be32(XLOG_HEADER_MAGIC_NUM)) {
                        found = 1;
                        break;
                }
        }
        /*
         * If we haven't found the log record header block, start looking
         * again from the end of the physical log.  XXXmiken: There should be
         * a check here to make sure we didn't search more than N blocks in
         * the previous code.
         */
        if (!found) {
                for (i = log->l_logBBsize - 1; i >= (int)(*head_blk); i--) {
                        error = xlog_bread(log, i, 1, bp, &offset);
                        if (error)
                                goto done;

                        if (*(__be32 *)offset ==
                            cpu_to_be32(XLOG_HEADER_MAGIC_NUM)) {
                                found = 2;
                                break;
                        }
                }
        }
        if (!found) {
                xfs_warn(log->l_mp, "%s: couldn't find sync record", __func__);
                ASSERT(0);
                return XFS_ERROR(EIO);
        }

        /* find blk_no of tail of log */
        rhead = (xlog_rec_header_t *)offset;
        *tail_blk = BLOCK_LSN(be64_to_cpu(rhead->h_tail_lsn));

        /*
         * Reset log values according to the state of the log when we
         * crashed.  In the case where head_blk == 0, we bump curr_cycle
         * one because the next write starts a new cycle rather than
         * continuing the cycle of the last good log record.  At this
         * point we have guaranteed that all partial log records have been
         * accounted for.  Therefore, we know that the last good log record
         * written was complete and ended exactly on the end boundary
         * of the physical log.
         */
        log->l_prev_block = i;
        log->l_curr_block = (int)*head_blk;
        log->l_curr_cycle = be32_to_cpu(rhead->h_cycle);
        if (found == 2)
                log->l_curr_cycle++;
        atomic64_set(&log->l_tail_lsn, be64_to_cpu(rhead->h_tail_lsn));
        atomic64_set(&log->l_last_sync_lsn, be64_to_cpu(rhead->h_lsn));
        xlog_assign_grant_head(&log->l_reserve_head.grant, log->l_curr_cycle,
                                        BBTOB(log->l_curr_block));
        xlog_assign_grant_head(&log->l_write_head.grant, log->l_curr_cycle,
                                        BBTOB(log->l_curr_block));

        /*
         * Look for unmount record.  If we find it, then we know there
         * was a clean unmount.  Since 'i' could be the last block in
         * the physical log, we convert to a log block before comparing
         * to the head_blk.
         *
         * Save the current tail lsn to use to pass to
         * xlog_clear_stale_blocks() below.  We won't want to clear the
         * unmount record if there is one, so we pass the lsn of the
         * unmount record rather than the block after it.
         */
        if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
                int     h_size = be32_to_cpu(rhead->h_size);
                int     h_version = be32_to_cpu(rhead->h_version);

                if ((h_version & XLOG_VERSION_2) &&
                    (h_size > XLOG_HEADER_CYCLE_SIZE)) {

                        hblks = h_size / XLOG_HEADER_CYCLE_SIZE;
                        if (h_size % XLOG_HEADER_CYCLE_SIZE)
                                hblks++;
                } else {
                        hblks = 1;
                }
        } else {
                hblks = 1;
        }
        after_umount_blk = (i + hblks + (int)
                BTOBB(be32_to_cpu(rhead->h_len))) % log->l_logBBsize;
        tail_lsn = atomic64_read(&log->l_tail_lsn);
        if (*head_blk == after_umount_blk &&
            be32_to_cpu(rhead->h_num_logops) == 1) {
                umount_data_blk = (i + hblks) % log->l_logBBsize;
                error = xlog_bread(log, umount_data_blk, 1, bp, &offset);
                if (error)
                        goto done;

                op_head = (xlog_op_header_t *)offset;
                if (op_head->oh_flags & XLOG_UNMOUNT_TRANS) {
                        /*
                         * Set tail and last sync so that newly written
                         * log records will point recovery to after the
                         * current unmount record.
                         */
                        xlog_assign_atomic_lsn(&log->l_tail_lsn,
                                        log->l_curr_cycle, after_umount_blk);
                        xlog_assign_atomic_lsn(&log->l_last_sync_lsn,
                                        log->l_curr_cycle, after_umount_blk);
                        *tail_blk = after_umount_blk;

                        /*
                         * Note that the unmount was clean. If the unmount
                         * was not clean, we need to know this to rebuild the
                         * superblock counters from the perag headers if we
                         * have a filesystem using non-persistent counters.
                         */
                        log->l_mp->m_flags |= XFS_MOUNT_WAS_CLEAN;
                }
        }

        /*
         * Make sure that there are no blocks in front of the head
         * with the same cycle number as the head.  This can happen
         * because we allow multiple outstanding log writes concurrently,
         * and the later writes might make it out before earlier ones.
         *
         * We use the lsn from before modifying it so that we'll never
         * overwrite the unmount record after a clean unmount.
         *
         * Do this only if we are going to recover the filesystem
         *
         * NOTE: This used to say "if (!readonly)"
         * However on Linux, we can & do recover a read-only filesystem.
         * We only skip recovery if NORECOVERY is specified on mount,
         * in which case we would not be here.
         *
         * But... if the -device- itself is readonly, just skip this.
         * We can't recover this device anyway, so it won't matter.
         */
        if (!xfs_readonly_buftarg(log->l_mp->m_logdev_targp))
                error = xlog_clear_stale_blocks(log, tail_lsn);

done:
        xlog_put_bp(bp);

        if (error)
                xfs_warn(log->l_mp, "failed to locate log tail");
        return error;
}

/*
 * Is the log zeroed at all?
 *
 * The last binary search should be changed to perform an X block read
 * once X becomes small enough.  You can then search linearly through
 * the X blocks.  This will cut down on the number of reads we need to do.
 *
 * If the log is partially zeroed, this routine will pass back the blkno
 * of the first block with cycle number 0.  It won't have a complete LR
 * preceding it.
 *
 * Return:
 *      0  => the log is completely written to
 *      -1 => use *blk_no as the first block of the log
 *      >0 => error has occurred
 */
STATIC int
xlog_find_zeroed(
        struct xlog     *log,
        xfs_daddr_t     *blk_no)
{
        xfs_buf_t       *bp;
        xfs_caddr_t     offset;
        uint            first_cycle, last_cycle;
        xfs_daddr_t     new_blk, last_blk, start_blk;
        xfs_daddr_t     num_scan_bblks;
        int             error, log_bbnum = log->l_logBBsize;

        *blk_no = 0;

        /* check totally zeroed log */
        bp = xlog_get_bp(log, 1);
        if (!bp)
                return ENOMEM;
        error = xlog_bread(log, 0, 1, bp, &offset);
        if (error)
                goto bp_err;

        first_cycle = xlog_get_cycle(offset);
        if (first_cycle == 0) {         /* completely zeroed log */
                *blk_no = 0;
                xlog_put_bp(bp);
                return -1;
        }

        /* check partially zeroed log */
        error = xlog_bread(log, log_bbnum-1, 1, bp, &offset);
        if (error)
                goto bp_err;

        last_cycle = xlog_get_cycle(offset);
        if (last_cycle != 0) {          /* log completely written to */
                xlog_put_bp(bp);
                return 0;
        } else if (first_cycle != 1) {
                /*
                 * If the cycle of the last block is zero, the cycle of
                 * the first block must be 1. If it's not, maybe we're
                 * not looking at a log... Bail out.
                 */
                xfs_warn(log->l_mp,
                        "Log inconsistent or not a log (last==0, first!=1)");
                return XFS_ERROR(EINVAL);
        }

        /* we have a partially zeroed log */
        last_blk = log_bbnum-1;
        if ((error = xlog_find_cycle_start(log, bp, 0, &last_blk, 0)))
                goto bp_err;

        /*
         * Validate the answer.  Because there is no way to guarantee that
         * the entire log is made up of log records which are the same size,
         * we scan over the defined maximum blocks.  At this point, the maximum
         * is not chosen to mean anything special.   XXXmiken
         */
        num_scan_bblks = XLOG_TOTAL_REC_SHIFT(log);
        ASSERT(num_scan_bblks <= INT_MAX);

        if (last_blk < num_scan_bblks)
                num_scan_bblks = last_blk;
        start_blk = last_blk - num_scan_bblks;

        /*
         * We search for any instances of cycle number 0 that occur before
         * our current estimate of the head.  What we're trying to detect is
         *        1 ... | 0 | 1 | 0...
         *                       ^ binary search ends here
         */
        if ((error = xlog_find_verify_cycle(log, start_blk,
                                         (int)num_scan_bblks, 0, &new_blk)))
                goto bp_err;
        if (new_blk != -1)
                last_blk = new_blk;

        /*
         * Potentially backup over partial log record write.  We don't need
         * to search the end of the log because we know it is zero.
         */
        if ((error = xlog_find_verify_log_record(log, start_blk,
                                &last_blk, 0)) == -1) {
            error = XFS_ERROR(EIO);
            goto bp_err;
        } else if (error)
            goto bp_err;

        *blk_no = last_blk;
bp_err:
        xlog_put_bp(bp);
        if (error)
                return error;
        return -1;
}

/*
 * These are simple subroutines used by xlog_clear_stale_blocks() below
 * to initialize a buffer full of empty log record headers and write
 * them into the log.
 */
STATIC void
xlog_add_record(
        struct xlog             *log,
        xfs_caddr_t             buf,
        int                     cycle,
        int                     block,
        int                     tail_cycle,
        int                     tail_block)
{
        xlog_rec_header_t       *recp = (xlog_rec_header_t *)buf;

        memset(buf, 0, BBSIZE);
        recp->h_magicno = cpu_to_be32(XLOG_HEADER_MAGIC_NUM);
        recp->h_cycle = cpu_to_be32(cycle);
        recp->h_version = cpu_to_be32(
                        xfs_sb_version_haslogv2(&log->l_mp->m_sb) ? 2 : 1);
        recp->h_lsn = cpu_to_be64(xlog_assign_lsn(cycle, block));
        recp->h_tail_lsn = cpu_to_be64(xlog_assign_lsn(tail_cycle, tail_block));
        recp->h_fmt = cpu_to_be32(XLOG_FMT);
        memcpy(&recp->h_fs_uuid, &log->l_mp->m_sb.sb_uuid, sizeof(uuid_t));
}

STATIC int
xlog_write_log_records(
        struct xlog     *log,
        int             cycle,
        int             start_block,
        int             blocks,
        int             tail_cycle,
        int             tail_block)
{
        xfs_caddr_t     offset;
        xfs_buf_t       *bp;
        int             balign, ealign;
        int             sectbb = log->l_sectBBsize;
        int             end_block = start_block + blocks;
        int             bufblks;
        int             error = 0;
        int             i, j = 0;

        /*
         * Greedily allocate a buffer big enough to handle the full
         * range of basic blocks to be written.  If that fails, try
         * a smaller size.  We need to be able to write at least a
         * log sector, or we're out of luck.
         */
        bufblks = 1 << ffs(blocks);
        while (bufblks > log->l_logBBsize)
                bufblks >>= 1;
        while (!(bp = xlog_get_bp(log, bufblks))) {
                bufblks >>= 1;
                if (bufblks < sectbb)
                        return ENOMEM;
        }

        /* We may need to do a read at the start to fill in part of
         * the buffer in the starting sector not covered by the first
         * write below.
         */
        balign = round_down(start_block, sectbb);
        if (balign != start_block) {
                error = xlog_bread_noalign(log, start_block, 1, bp);
                if (error)
                        goto out_put_bp;

                j = start_block - balign;
        }

        for (i = start_block; i < end_block; i += bufblks) {
                int             bcount, endcount;

                bcount = min(bufblks, end_block - start_block);
                endcount = bcount - j;

                /* We may need to do a read at the end to fill in part of
                 * the buffer in the final sector not covered by the write.
                 * If this is the same sector as the above read, skip it.
                 */
                ealign = round_down(end_block, sectbb);
                if (j == 0 && (start_block + endcount > ealign)) {
                        offset = bp->b_addr + BBTOB(ealign - start_block);
                        error = xlog_bread_offset(log, ealign, sectbb,
                                                        bp, offset);
                        if (error)
                                break;

                }

                offset = xlog_align(log, start_block, endcount, bp);
                for (; j < endcount; j++) {
                        xlog_add_record(log, offset, cycle, i+j,
                                        tail_cycle, tail_block);
                        offset += BBSIZE;
                }
                error = xlog_bwrite(log, start_block, endcount, bp);
                if (error)
                        break;
                start_block += endcount;
                j = 0;
        }

 out_put_bp:
        xlog_put_bp(bp);
        return error;
}

/*
 * This routine is called to blow away any incomplete log writes out
 * in front of the log head.  We do this so that we won't become confused
 * if we come up, write only a little bit more, and then crash again.
 * If we leave the partial log records out there, this situation could
 * cause us to think those partial writes are valid blocks since they
 * have the current cycle number.  We get rid of them by overwriting them
 * with empty log records with the old cycle number rather than the
 * current one.
 *
 * The tail lsn is passed in rather than taken from
 * the log so that we will not write over the unmount record after a
 * clean unmount in a 512 block log.  Doing so would leave the log without
 * any valid log records in it until a new one was written.  If we crashed
 * during that time we would not be able to recover.
 */
STATIC int
xlog_clear_stale_blocks(
        struct xlog     *log,
        xfs_lsn_t       tail_lsn)
{
        int             tail_cycle, head_cycle;
        int             tail_block, head_block;
        int             tail_distance, max_distance;
        int             distance;
        int             error;

        tail_cycle = CYCLE_LSN(tail_lsn);
        tail_block = BLOCK_LSN(tail_lsn);
        head_cycle = log->l_curr_cycle;
        head_block = log->l_curr_block;

        /*
         * Figure out the distance between the new head of the log
         * and the tail.  We want to write over any blocks beyond the
         * head that we may have written just before the crash, but
         * we don't want to overwrite the tail of the log.
         */
        if (head_cycle == tail_cycle) {
                /*
                 * The tail is behind the head in the physical log,
                 * so the distance from the head to the tail is the
                 * distance from the head to the end of the log plus
                 * the distance from the beginning of the log to the
                 * tail.
                 */
                if (unlikely(head_block < tail_block || head_block >= log->l_logBBsize)) {
                        XFS_ERROR_REPORT("xlog_clear_stale_blocks(1)",
                                         XFS_ERRLEVEL_LOW, log->l_mp);
                        return XFS_ERROR(EFSCORRUPTED);
                }
                tail_distance = tail_block + (log->l_logBBsize - head_block);
        } else {
                /*
                 * The head is behind the tail in the physical log,
                 * so the distance from the head to the tail is just
                 * the tail block minus the head block.
                 */
                if (unlikely(head_block >= tail_block || head_cycle != (tail_cycle + 1))){
                        XFS_ERROR_REPORT("xlog_clear_stale_blocks(2)",
                                         XFS_ERRLEVEL_LOW, log->l_mp);
                        return XFS_ERROR(EFSCORRUPTED);
                }
                tail_distance = tail_block - head_block;
        }

        /*
         * If the head is right up against the tail, we can't clear
         * anything.
         */
        if (tail_distance <= 0) {
                ASSERT(tail_distance == 0);
                return 0;
        }

        max_distance = XLOG_TOTAL_REC_SHIFT(log);
        /*
         * Take the smaller of the maximum amount of outstanding I/O
         * we could have and the distance to the tail to clear out.
         * We take the smaller so that we don't overwrite the tail and
         * we don't waste all day writing from the head to the tail
         * for no reason.
         */
        max_distance = MIN(max_distance, tail_distance);

        if ((head_block + max_distance) <= log->l_logBBsize) {
                /*
                 * We can stomp all the blocks we need to without
                 * wrapping around the end of the log.  Just do it
                 * in a single write.  Use the cycle number of the
                 * current cycle minus one so that the log will look like:
                 *     n ... | n - 1 ...
                 */
                error = xlog_write_log_records(log, (head_cycle - 1),
                                head_block, max_distance, tail_cycle,
                                tail_block);
                if (error)
                        return error;
        } else {
                /*
                 * We need to wrap around the end of the physical log in
                 * order to clear all the blocks.  Do it in two separate
                 * I/Os.  The first write should be from the head to the
                 * end of the physical log, and it should use the current
                 * cycle number minus one just like above.
                 */
                distance = log->l_logBBsize - head_block;
                error = xlog_write_log_records(log, (head_cycle - 1),
                                head_block, distance, tail_cycle,
                                tail_block);

                if (error)
                        return error;

                /*
                 * Now write the blocks at the start of the physical log.
                 * This writes the remainder of the blocks we want to clear.
                 * It uses the current cycle number since we're now on the
                 * same cycle as the head so that we get:
                 *    n ... n ... | n - 1 ...
                 *    ^^^^^ blocks we're writing
                 */
                distance = max_distance - (log->l_logBBsize - head_block);
                error = xlog_write_log_records(log, head_cycle, 0, distance,
                                tail_cycle, tail_block);
                if (error)
                        return error;
        }

        return 0;
}

/******************************************************************************
 *
 *              Log recover routines
 *
 ******************************************************************************
 */

STATIC xlog_recover_t *
xlog_recover_find_tid(
        struct hlist_head       *head,
        xlog_tid_t              tid)
{
        xlog_recover_t          *trans;
        struct hlist_node       *n;

        hlist_for_each_entry(trans, n, head, r_list) {
                if (trans->r_log_tid == tid)
                        return trans;
        }
        return NULL;
}

STATIC void
xlog_recover_new_tid(
        struct hlist_head       *head,
        xlog_tid_t              tid,
        xfs_lsn_t               lsn)
{
        xlog_recover_t          *trans;

        trans = kmem_zalloc(sizeof(xlog_recover_t), KM_SLEEP);
        trans->r_log_tid   = tid;
        trans->r_lsn       = lsn;
        INIT_LIST_HEAD(&trans->r_itemq);

        INIT_HLIST_NODE(&trans->r_list);
        hlist_add_head(&trans->r_list, head);
}

STATIC void
xlog_recover_add_item(
        struct list_head        *head)
{
        xlog_recover_item_t     *item;

        item = kmem_zalloc(sizeof(xlog_recover_item_t), KM_SLEEP);
        INIT_LIST_HEAD(&item->ri_list);
        list_add_tail(&item->ri_list, head);
}

STATIC int
xlog_recover_add_to_cont_trans(
        struct xlog             *log,
        struct xlog_recover     *trans,
        xfs_caddr_t             dp,
        int                     len)
{
        xlog_recover_item_t     *item;
        xfs_caddr_t             ptr, old_ptr;
        int                     old_len;

        if (list_empty(&trans->r_itemq)) {
                /* finish copying rest of trans header */
                xlog_recover_add_item(&trans->r_itemq);
                ptr = (xfs_caddr_t) &trans->r_theader +
                                sizeof(xfs_trans_header_t) - len;
                memcpy(ptr, dp, len); /* d, s, l */
                return 0;
        }
        /* take the tail entry */
        item = list_entry(trans->r_itemq.prev, xlog_recover_item_t, ri_list);

        old_ptr = item->ri_buf[item->ri_cnt-1].i_addr;
        old_len = item->ri_buf[item->ri_cnt-1].i_len;

        ptr = kmem_realloc(old_ptr, len+old_len, old_len, KM_SLEEP);
        memcpy(&ptr[old_len], dp, len); /* d, s, l */
        item->ri_buf[item->ri_cnt-1].i_len += len;
        item->ri_buf[item->ri_cnt-1].i_addr = ptr;
        trace_xfs_log_recover_item_add_cont(log, trans, item, 0);
        return 0;
}

/*
 * The next region to add is the start of a new region.  It could be
 * a whole region or it could be the first part of a new region.  Because
 * of this, the assumption here is that the type and size fields of all
 * format structures fit into the first 32 bits of the structure.
 *
 * This works because all regions must be 32 bit aligned.  Therefore, we
 * either have both fields or we have neither field.  In the case we have
 * neither field, the data part of the region is zero length.  We only have
 * a log_op_header and can throw away the header since a new one will appear
 * later.  If we have at least 4 bytes, then we can determine how many regions
 * will appear in the current log item.
 */
STATIC int
xlog_recover_add_to_trans(
        struct xlog             *log,
        struct xlog_recover     *trans,
        xfs_caddr_t             dp,
        int                     len)
{
        xfs_inode_log_format_t  *in_f;                  /* any will do */
        xlog_recover_item_t     *item;
        xfs_caddr_t             ptr;

        if (!len)
                return 0;
        if (list_empty(&trans->r_itemq)) {
                /* we need to catch log corruptions here */
                if (*(uint *)dp != XFS_TRANS_HEADER_MAGIC) {
                        xfs_warn(log->l_mp, "%s: bad header magic number",
                                __func__);
                        ASSERT(0);
                        return XFS_ERROR(EIO);
                }
                if (len == sizeof(xfs_trans_header_t))
                        xlog_recover_add_item(&trans->r_itemq);
                memcpy(&trans->r_theader, dp, len); /* d, s, l */
                return 0;
        }

        ptr = kmem_alloc(len, KM_SLEEP);
        memcpy(ptr, dp, len);
        in_f = (xfs_inode_log_format_t *)ptr;

        /* take the tail entry */
        item = list_entry(trans->r_itemq.prev, xlog_recover_item_t, ri_list);
        if (item->ri_total != 0 &&
             item->ri_total == item->ri_cnt) {
                /* tail item is in use, get a new one */
                xlog_recover_add_item(&trans->r_itemq);
                item = list_entry(trans->r_itemq.prev,
                                        xlog_recover_item_t, ri_list);
        }

        if (item->ri_total == 0) {              /* first region to be added */
                if (in_f->ilf_size == 0 ||
                    in_f->ilf_size > XLOG_MAX_REGIONS_IN_ITEM) {
                        xfs_warn(log->l_mp,
                "bad number of regions (%d) in inode log format",
                                  in_f->ilf_size);
                        ASSERT(0);
                        return XFS_ERROR(EIO);
                }

                item->ri_total = in_f->ilf_size;
                item->ri_buf =
                        kmem_zalloc(item->ri_total * sizeof(xfs_log_iovec_t),
                                    KM_SLEEP);
        }
        ASSERT(item->ri_total > item->ri_cnt);
        /* Description region is ri_buf[0] */
        item->ri_buf[item->ri_cnt].i_addr = ptr;
        item->ri_buf[item->ri_cnt].i_len  = len;
        item->ri_cnt++;
        trace_xfs_log_recover_item_add(log, trans, item, 0);
        return 0;
}

/*
 * Sort the log items in the transaction. Cancelled buffers need
 * to be put first so they are processed before any items that might
 * modify the buffers. If they are cancelled, then the modifications
 * don't need to be replayed.
 */
STATIC int
xlog_recover_reorder_trans(
        struct xlog             *log,
        struct xlog_recover     *trans,
        int                     pass)
{
        xlog_recover_item_t     *item, *n;
        LIST_HEAD(sort_list);

        list_splice_init(&trans->r_itemq, &sort_list);
        list_for_each_entry_safe(item, n, &sort_list, ri_list) {
                xfs_buf_log_format_t    *buf_f = item->ri_buf[0].i_addr;

                switch (ITEM_TYPE(item)) {
                case XFS_LI_BUF:
                        if (!(buf_f->blf_flags & XFS_BLF_CANCEL)) {
                                trace_xfs_log_recover_item_reorder_head(log,
                                                        trans, item, pass);
                                list_move(&item->ri_list, &trans->r_itemq);
                                break;
                        }
                case XFS_LI_INODE:
                case XFS_LI_DQUOT:
                case XFS_LI_QUOTAOFF:
                case XFS_LI_EFD:
                case XFS_LI_EFI:
                        trace_xfs_log_recover_item_reorder_tail(log,
                                                        trans, item, pass);
                        list_move_tail(&item->ri_list, &trans->r_itemq);
                        break;
                default:
                        xfs_warn(log->l_mp,
                                "%s: unrecognized type of log operation",
                                __func__);
                        ASSERT(0);
                        return XFS_ERROR(EIO);
                }
        }
        ASSERT(list_empty(&sort_list));
        return 0;
}

/*
 * Build up the table of buf cancel records so that we don't replay
 * cancelled data in the second pass.  For buffer records that are
 * not cancel records, there is nothing to do here so we just return.
 *
 * If we get a cancel record which is already in the table, this indicates
 * that the buffer was cancelled multiple times.  In order to ensure
 * that during pass 2 we keep the record in the table until we reach its
 * last occurrence in the log, we keep a reference count in the cancel
 * record in the table to tell us how many times we expect to see this
 * record during the second pass.
 */
STATIC int
xlog_recover_buffer_pass1(
        struct xlog                     *log,
        struct xlog_recover_item        *item)
{
        xfs_buf_log_format_t    *buf_f = item->ri_buf[0].i_addr;
        struct list_head        *bucket;
        struct xfs_buf_cancel   *bcp;

        /*
         * If this isn't a cancel buffer item, then just return.
         */
        if (!(buf_f->blf_flags & XFS_BLF_CANCEL)) {
                trace_xfs_log_recover_buf_not_cancel(log, buf_f);
                return 0;
        }

        /*
         * Insert an xfs_buf_cancel record into the hash table of them.
         * If there is already an identical record, bump its reference count.
         */
        bucket = XLOG_BUF_CANCEL_BUCKET(log, buf_f->blf_blkno);
        list_for_each_entry(bcp, bucket, bc_list) {
                if (bcp->bc_blkno == buf_f->blf_blkno &&
                    bcp->bc_len == buf_f->blf_len) {
                        bcp->bc_refcount++;
                        trace_xfs_log_recover_buf_cancel_ref_inc(log, buf_f);
                        return 0;
                }
        }

        bcp = kmem_alloc(sizeof(struct xfs_buf_cancel), KM_SLEEP);
        bcp->bc_blkno = buf_f->blf_blkno;
        bcp->bc_len = buf_f->blf_len;
        bcp->bc_refcount = 1;
        list_add_tail(&bcp->bc_list, bucket);

        trace_xfs_log_recover_buf_cancel_add(log, buf_f);
        return 0;
}

/*
 * Check to see whether the buffer being recovered has a corresponding
 * entry in the buffer cancel record table.  If it does then return 1
 * so that it will be cancelled, otherwise return 0.  If the buffer is
 * actually a buffer cancel item (XFS_BLF_CANCEL is set), then decrement
 * the refcount on the entry in the table and remove it from the table
 * if this is the last reference.
 *
 * We remove the cancel record from the table when we encounter its
 * last occurrence in the log so that if the same buffer is re-used
 * again after its last cancellation we actually replay the changes
 * made at that point.
 */
STATIC int
xlog_check_buffer_cancelled(
        struct xlog             *log,
        xfs_daddr_t             blkno,
        uint                    len,
        ushort                  flags)
{
        struct list_head        *bucket;
        struct xfs_buf_cancel   *bcp;

        if (log->l_buf_cancel_table == NULL) {
                /*
                 * There is nothing in the table built in pass one,
                 * so this buffer must not be cancelled.
                 */
                ASSERT(!(flags & XFS_BLF_CANCEL));
                return 0;
        }

        /*
         * Search for an entry in the  cancel table that matches our buffer.
         */
        bucket = XLOG_BUF_CANCEL_BUCKET(log, blkno);
        list_for_each_entry(bcp, bucket, bc_list) {
                if (bcp->bc_blkno == blkno && bcp->bc_len == len)
                        goto found;
        }

        /*
         * We didn't find a corresponding entry in the table, so return 0 so
         * that the buffer is NOT cancelled.
         */
        ASSERT(!(flags & XFS_BLF_CANCEL));
        return 0;

found:
        /*
         * We've go a match, so return 1 so that the recovery of this buffer
         * is cancelled.  If this buffer is actually a buffer cancel log
         * item, then decrement the refcount on the one in the table and
         * remove it if this is the last reference.
         */
        if (flags & XFS_BLF_CANCEL) {
                if (--bcp->bc_refcount == 0) {
                        list_del(&bcp->bc_list);
                        kmem_free(bcp);
                }
        }
        return 1;
}

/*
 * Perform recovery for a buffer full of inodes.  In these buffers, the only
 * data which should be recovered is that which corresponds to the
 * di_next_unlinked pointers in the on disk inode structures.  The rest of the
 * data for the inodes is always logged through the inodes themselves rather
 * than the inode buffer and is recovered in xlog_recover_inode_pass2().
 *
 * The only time when buffers full of inodes are fully recovered is when the
 * buffer is full of newly allocated inodes.  In this case the buffer will
 * not be marked as an inode buffer and so will be sent to
 * xlog_recover_do_reg_buffer() below during recovery.
 */
STATIC int
xlog_recover_do_inode_buffer(
        struct xfs_mount        *mp,
        xlog_recover_item_t     *item,
        struct xfs_buf          *bp,
        xfs_buf_log_format_t    *buf_f)
{
        int                     i;
        int                     item_index = 0;
        int                     bit = 0;
        int                     nbits = 0;
        int                     reg_buf_offset = 0;
        int                     reg_buf_bytes = 0;
        int                     next_unlinked_offset;
        int                     inodes_per_buf;
        xfs_agino_t             *logged_nextp;
        xfs_agino_t             *buffer_nextp;

        trace_xfs_log_recover_buf_inode_buf(mp->m_log, buf_f);

        inodes_per_buf = BBTOB(bp->b_io_length) >> mp->m_sb.sb_inodelog;
        for (i = 0; i < inodes_per_buf; i++) {
                next_unlinked_offset = (i * mp->m_sb.sb_inodesize) +
                        offsetof(xfs_dinode_t, di_next_unlinked);

                while (next_unlinked_offset >=
                       (reg_buf_offset + reg_buf_bytes)) {
                        /*
                         * The next di_next_unlinked field is beyond
                         * the current logged region.  Find the next
                         * logged region that contains or is beyond
                         * the current di_next_unlinked field.
                         */
                        bit += nbits;
                        bit = xfs_next_bit(buf_f->blf_data_map,
                                           buf_f->blf_map_size, bit);

                        /*
                         * If there are no more logged regions in the
                         * buffer, then we're done.
                         */
                        if (bit == -1)
                                return 0;

                        nbits = xfs_contig_bits(buf_f->blf_data_map,
                                                buf_f->blf_map_size, bit);
                        ASSERT(nbits > 0);
                        reg_buf_offset = bit << XFS_BLF_SHIFT;
                        reg_buf_bytes = nbits << XFS_BLF_SHIFT;
                        item_index++;
                }

                /*
                 * If the current logged region starts after the current
                 * di_next_unlinked field, then move on to the next
                 * di_next_unlinked field.
                 */
                if (next_unlinked_offset < reg_buf_offset)
                        continue;

                ASSERT(item->ri_buf[item_index].i_addr != NULL);
                ASSERT((item->ri_buf[item_index].i_len % XFS_BLF_CHUNK) == 0);
                ASSERT((reg_buf_offset + reg_buf_bytes) <=
                                                        BBTOB(bp->b_io_length));

                /*
                 * The current logged region contains a copy of the
                 * current di_next_unlinked field.  Extract its value
                 * and copy it to the buffer copy.
                 */
                logged_nextp = item->ri_buf[item_index].i_addr +
                                next_unlinked_offset - reg_buf_offset;
                if (unlikely(*logged_nextp == 0)) {
                        xfs_alert(mp,
                "Bad inode buffer log record (ptr = 0x%p, bp = 0x%p). "
                "Trying to replay bad (0) inode di_next_unlinked field.",
                                item, bp);
                        XFS_ERROR_REPORT("xlog_recover_do_inode_buf",
                                         XFS_ERRLEVEL_LOW, mp);
                        return XFS_ERROR(EFSCORRUPTED);
                }

                buffer_nextp = (xfs_agino_t *)xfs_buf_offset(bp,
                                              next_unlinked_offset);
                *buffer_nextp = *logged_nextp;
        }

        return 0;
}

/*
 * Perform a 'normal' buffer recovery.  Each logged region of the
 * buffer should be copied over the corresponding region in the
 * given buffer.  The bitmap in the buf log format structure indicates
 * where to place the logged data.
 */
STATIC void
xlog_recover_do_reg_buffer(
        struct xfs_mount        *mp,
        xlog_recover_item_t     *item,
        struct xfs_buf          *bp,
        xfs_buf_log_format_t    *buf_f)
{
        int                     i;
        int                     bit;
        int                     nbits;
        int                     error;

        trace_xfs_log_recover_buf_reg_buf(mp->m_log, buf_f);

        bit = 0;
        i = 1;  /* 0 is the buf format structure */
        while (1) {
                bit = xfs_next_bit(buf_f->blf_data_map,
                                   buf_f->blf_map_size, bit);
                if (bit == -1)
                        break;
                nbits = xfs_contig_bits(buf_f->blf_data_map,
                                        buf_f->blf_map_size, bit);
                ASSERT(nbits > 0);
                ASSERT(item->ri_buf[i].i_addr != NULL);
                ASSERT(item->ri_buf[i].i_len % XFS_BLF_CHUNK == 0);
                ASSERT(BBTOB(bp->b_io_length) >=
                       ((uint)bit << XFS_BLF_SHIFT) + (nbits << XFS_BLF_SHIFT));

                /*
                 * Do a sanity check if this is a dquot buffer. Just checking
                 * the first dquot in the buffer should do. XXXThis is
                 * probably a good thing to do for other buf types also.
                 */
                error = 0;
                if (buf_f->blf_flags &
                   (XFS_BLF_UDQUOT_BUF|XFS_BLF_PDQUOT_BUF|XFS_BLF_GDQUOT_BUF)) {
                        if (item->ri_buf[i].i_addr == NULL) {
                                xfs_alert(mp,
                                        "XFS: NULL dquot in %s.", __func__);
                                goto next;
                        }
                        if (item->ri_buf[i].i_len < sizeof(xfs_disk_dquot_t)) {
                                xfs_alert(mp,
                                        "XFS: dquot too small (%d) in %s.",
                                        item->ri_buf[i].i_len, __func__);
                                goto next;
                        }
                        error = xfs_qm_dqcheck(mp, item->ri_buf[i].i_addr,
                                               -1, 0, XFS_QMOPT_DOWARN,
                                               "dquot_buf_recover");
                        if (error)
                                goto next;
                }

                memcpy(xfs_buf_offset(bp,
                        (uint)bit << XFS_BLF_SHIFT),    /* dest */
                        item->ri_buf[i].i_addr,         /* source */
                        nbits<<XFS_BLF_SHIFT);          /* length */
 next:
                i++;
                bit += nbits;
        }

        /* Shouldn't be any more regions */
        ASSERT(i == item->ri_total);
}

/*
 * Do some primitive error checking on ondisk dquot data structures.
 */
int
xfs_qm_dqcheck(
        struct xfs_mount *mp,
        xfs_disk_dquot_t *ddq,
        xfs_dqid_t       id,
        uint             type,    /* used only when IO_dorepair is true */
        uint             flags,
        char             *str)
{
        xfs_dqblk_t      *d = (xfs_dqblk_t *)ddq;
        int             errs = 0;

        /*
         * We can encounter an uninitialized dquot buffer for 2 reasons:
         * 1. If we crash while deleting the quotainode(s), and those blks got
         *    used for user data. This is because we take the path of regular
         *    file deletion; however, the size field of quotainodes is never
         *    updated, so all the tricks that we play in itruncate_finish
         *    don't quite matter.
         *
         * 2. We don't play the quota buffers when there's a quotaoff logitem.
         *    But the allocation will be replayed so we'll end up with an
         *    uninitialized quota block.
         *
         * This is all fine; things are still consistent, and we haven't lost
         * any quota information. Just don't complain about bad dquot blks.
         */
        if (ddq->d_magic != cpu_to_be16(XFS_DQUOT_MAGIC)) {
                if (flags & XFS_QMOPT_DOWARN)
                        xfs_alert(mp,
                        "%s : XFS dquot ID 0x%x, magic 0x%x != 0x%x",
                        str, id, be16_to_cpu(ddq->d_magic), XFS_DQUOT_MAGIC);
                errs++;
        }
        if (ddq->d_version != XFS_DQUOT_VERSION) {
                if (flags & XFS_QMOPT_DOWARN)
                        xfs_alert(mp,
                        "%s : XFS dquot ID 0x%x, version 0x%x != 0x%x",
                        str, id, ddq->d_version, XFS_DQUOT_VERSION);
                errs++;
        }

        if (ddq->d_flags != XFS_DQ_USER &&
            ddq->d_flags != XFS_DQ_PROJ &&
            ddq->d_flags != XFS_DQ_GROUP) {
                if (flags & XFS_QMOPT_DOWARN)
                        xfs_alert(mp,
                        "%s : XFS dquot ID 0x%x, unknown flags 0x%x",
                        str, id, ddq->d_flags);
                errs++;
        }

        if (id != -1 && id != be32_to_cpu(ddq->d_id)) {
                if (flags & XFS_QMOPT_DOWARN)
                        xfs_alert(mp,
                        "%s : ondisk-dquot 0x%p, ID mismatch: "
                        "0x%x expected, found id 0x%x",
                        str, ddq, id, be32_to_cpu(ddq->d_id));
                errs++;
        }

        if (!errs && ddq->d_id) {
                if (ddq->d_blk_softlimit &&
                    be64_to_cpu(ddq->d_bcount) >
                                be64_to_cpu(ddq->d_blk_softlimit)) {
                        if (!ddq->d_btimer) {

                                if (flags & XFS_QMOPT_DOWARN)
                                        xfs_alert(mp,
                        "%s : Dquot ID 0x%x (0x%p) BLK TIMER NOT STARTED",
                                        str, (int)be32_to_cpu(ddq->d_id), ddq);
                                errs++;
                        }
                }
                if (ddq->d_ino_softlimit &&
                    be64_to_cpu(ddq->d_icount) >
                                be64_to_cpu(ddq->d_ino_softlimit)) {
                        if (!ddq->d_itimer) {
                                if (flags & XFS_QMOPT_DOWARN)
                                        xfs_alert(mp,
                        "%s : Dquot ID 0x%x (0x%p) INODE TIMER NOT STARTED",
                                        str, (int)be32_to_cpu(ddq->d_id), ddq);
                                errs++;
                        }
                }
                if (ddq->d_rtb_softlimit &&
                    be64_to_cpu(ddq->d_rtbcount) >
                                be64_to_cpu(ddq->d_rtb_softlimit)) {
                        if (!ddq->d_rtbtimer) {
                                if (flags & XFS_QMOPT_DOWARN)
                                        xfs_alert(mp,
                        "%s : Dquot ID 0x%x (0x%p) RTBLK TIMER NOT STARTED",
                                        str, (int)be32_to_cpu(ddq->d_id), ddq);
                                errs++;
                        }
                }
        }

        if (!errs || !(flags & XFS_QMOPT_DQREPAIR))
                return errs;

        if (flags & XFS_QMOPT_DOWARN)
                xfs_notice(mp, "Re-initializing dquot ID 0x%x", id);

        /*
         * Typically, a repair is only requested by quotacheck.
         */
        ASSERT(id != -1);
        ASSERT(flags & XFS_QMOPT_DQREPAIR);
        memset(d, 0, sizeof(xfs_dqblk_t));

        d->dd_diskdq.d_magic = cpu_to_be16(XFS_DQUOT_MAGIC);
        d->dd_diskdq.d_version = XFS_DQUOT_VERSION;
        d->dd_diskdq.d_flags = type;
        d->dd_diskdq.d_id = cpu_to_be32(id);

        return errs;
}

/*
 * Perform a dquot buffer recovery.
 * Simple algorithm: if we have found a QUOTAOFF logitem of the same type
 * (ie. USR or GRP), then just toss this buffer away; don't recover it.
 * Else, treat it as a regular buffer and do recovery.
 */
STATIC void
xlog_recover_do_dquot_buffer(
        struct xfs_mount                *mp,
        struct xlog                     *log,
        struct xlog_recover_item        *item,
        struct xfs_buf                  *bp,
        struct xfs_buf_log_format       *buf_f)
{
        uint                    type;

        trace_xfs_log_recover_buf_dquot_buf(log, buf_f);

        /*
         * Filesystems are required to send in quota flags at mount time.
         */
        if (mp->m_qflags == 0) {
                return;
        }

        type = 0;
        if (buf_f->blf_flags & XFS_BLF_UDQUOT_BUF)
                type |= XFS_DQ_USER;
        if (buf_f->blf_flags & XFS_BLF_PDQUOT_BUF)
                type |= XFS_DQ_PROJ;
        if (buf_f->blf_flags & XFS_BLF_GDQUOT_BUF)
                type |= XFS_DQ_GROUP;
        /*
         * This type of quotas was turned off, so ignore this buffer
         */
        if (log->l_quotaoffs_flag & type)
                return;

        xlog_recover_do_reg_buffer(mp, item, bp, buf_f);
}

/*
 * This routine replays a modification made to a buffer at runtime.
 * There are actually two types of buffer, regular and inode, which
 * are handled differently.  Inode buffers are handled differently
 * in that we only recover a specific set of data from them, namely
 * the inode di_next_unlinked fields.  This is because all other inode
 * data is actually logged via inode records and any data we replay
 * here which overlaps that may be stale.
 *
 * When meta-data buffers are freed at run time we log a buffer item
 * with the XFS_BLF_CANCEL bit set to indicate that previous copies
 * of the buffer in the log should not be replayed at recovery time.
 * This is so that if the blocks covered by the buffer are reused for
 * file data before we crash we don't end up replaying old, freed
 * meta-data into a user's file.
 *
 * To handle the cancellation of buffer log items, we make two passes
 * over the log during recovery.  During the first we build a table of
 * those buffers which have been cancelled, and during the second we
 * only replay those buffers which do not have corresponding cancel
 * records in the table.  See xlog_recover_do_buffer_pass[1,2] above
 * for more details on the implementation of the table of cancel records.
 */
STATIC int
xlog_recover_buffer_pass2(
        struct xlog                     *log,
        struct list_head                *buffer_list,
        struct xlog_recover_item        *item)
{
        xfs_buf_log_format_t    *buf_f = item->ri_buf[0].i_addr;
        xfs_mount_t             *mp = log->l_mp;
        xfs_buf_t               *bp;
        int                     error;
        uint                    buf_flags;

        /*
         * In this pass we only want to recover all the buffers which have
         * not been cancelled and are not cancellation buffers themselves.
         */
        if (xlog_check_buffer_cancelled(log, buf_f->blf_blkno,
                        buf_f->blf_len, buf_f->blf_flags)) {
                trace_xfs_log_recover_buf_cancel(log, buf_f);
                return 0;
        }

        trace_xfs_log_recover_buf_recover(log, buf_f);

        buf_flags = 0;
        if (buf_f->blf_flags & XFS_BLF_INODE_BUF)
                buf_flags |= XBF_UNMAPPED;

        bp = xfs_buf_read(mp->m_ddev_targp, buf_f->blf_blkno, buf_f->blf_len,
                          buf_flags);
        if (!bp)
                return XFS_ERROR(ENOMEM);
        error = bp->b_error;
        if (error) {
                xfs_buf_ioerror_alert(bp, "xlog_recover_do..(read#1)");
                xfs_buf_relse(bp);
                return error;
        }

        if (buf_f->blf_flags & XFS_BLF_INODE_BUF) {
                error = xlog_recover_do_inode_buffer(mp, item, bp, buf_f);
        } else if (buf_f->blf_flags &
                  (XFS_BLF_UDQUOT_BUF|XFS_BLF_PDQUOT_BUF|XFS_BLF_GDQUOT_BUF)) {
                xlog_recover_do_dquot_buffer(mp, log, item, bp, buf_f);
        } else {
                xlog_recover_do_reg_buffer(mp, item, bp, buf_f);
        }
        if (error)
                return XFS_ERROR(error);

        /*
         * Perform delayed write on the buffer.  Asynchronous writes will be
         * slower when taking into account all the buffers to be flushed.
         *
         * Also make sure that only inode buffers with good sizes stay in
         * the buffer cache.  The kernel moves inodes in buffers of 1 block
         * or XFS_INODE_CLUSTER_SIZE bytes, whichever is bigger.  The inode
         * buffers in the log can be a different size if the log was generated
         * by an older kernel using unclustered inode buffers or a newer kernel
         * running with a different inode cluster size.  Regardless, if the
         * the inode buffer size isn't MAX(blocksize, XFS_INODE_CLUSTER_SIZE)
         * for *our* value of XFS_INODE_CLUSTER_SIZE, then we need to keep
         * the buffer out of the buffer cache so that the buffer won't
         * overlap with future reads of those inodes.
         */
        if (XFS_DINODE_MAGIC ==
            be16_to_cpu(*((__be16 *)xfs_buf_offset(bp, 0))) &&
            (BBTOB(bp->b_io_length) != MAX(log->l_mp->m_sb.sb_blocksize,
                        (__uint32_t)XFS_INODE_CLUSTER_SIZE(log->l_mp)))) {
                xfs_buf_stale(bp);
                error = xfs_bwrite(bp);
        } else {
                ASSERT(bp->b_target->bt_mount == mp);
                bp->b_iodone = xlog_recover_iodone;
                xfs_buf_delwri_queue(bp, buffer_list);
        }

        xfs_buf_relse(bp);
        return error;
}

STATIC int
xlog_recover_inode_pass2(
        struct xlog                     *log,
        struct list_head                *buffer_list,
        struct xlog_recover_item        *item)
{
        xfs_inode_log_format_t  *in_f;
        xfs_mount_t             *mp = log->l_mp;
        xfs_buf_t               *bp;
        xfs_dinode_t            *dip;
        int                     len;
        xfs_caddr_t             src;
        xfs_caddr_t             dest;
        int                     error;
        int                     attr_index;
        uint                    fields;
        xfs_icdinode_t          *dicp;
        int                     need_free = 0;

        if (item->ri_buf[0].i_len == sizeof(xfs_inode_log_format_t)) {
                in_f = item->ri_buf[0].i_addr;
        } else {
                in_f = kmem_alloc(sizeof(xfs_inode_log_format_t), KM_SLEEP);
                need_free = 1;
                error = xfs_inode_item_format_convert(&item->ri_buf[0], in_f);
                if (error)
                        goto error;
        }

        /*
         * Inode buffers can be freed, look out for it,
         * and do not replay the inode.
         */
        if (xlog_check_buffer_cancelled(log, in_f->ilf_blkno,
                                        in_f->ilf_len, 0)) {
                error = 0;
                trace_xfs_log_recover_inode_cancel(log, in_f);
                goto error;
        }
        trace_xfs_log_recover_inode_recover(log, in_f);

        bp = xfs_buf_read(mp->m_ddev_targp, in_f->ilf_blkno, in_f->ilf_len, 0);
        if (!bp) {
                error = ENOMEM;
                goto error;
        }
        error = bp->b_error;
        if (error) {
                xfs_buf_ioerror_alert(bp, "xlog_recover_do..(read#2)");
                xfs_buf_relse(bp);
                goto error;
        }
        ASSERT(in_f->ilf_fields & XFS_ILOG_CORE);
        dip = (xfs_dinode_t *)xfs_buf_offset(bp, in_f->ilf_boffset);

        /*
         * Make sure the place we're flushing out to really looks
         * like an inode!
         */
        if (unlikely(dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC))) {
                xfs_buf_relse(bp);
                xfs_alert(mp,
        "%s: Bad inode magic number, dip = 0x%p, dino bp = 0x%p, ino = %Ld",
                        __func__, dip, bp, in_f->ilf_ino);
                XFS_ERROR_REPORT("xlog_recover_inode_pass2(1)",
                                 XFS_ERRLEVEL_LOW, mp);
                error = EFSCORRUPTED;
                goto error;
        }
        dicp = item->ri_buf[1].i_addr;
        if (unlikely(dicp->di_magic != XFS_DINODE_MAGIC)) {
                xfs_buf_relse(bp);
                xfs_alert(mp,
                        "%s: Bad inode log record, rec ptr 0x%p, ino %Ld",
                        __func__, item, in_f->ilf_ino);
                XFS_ERROR_REPORT("xlog_recover_inode_pass2(2)",
                                 XFS_ERRLEVEL_LOW, mp);
                error = EFSCORRUPTED;
                goto error;
        }

        /* Skip replay when the on disk inode is newer than the log one */
        if (dicp->di_flushiter < be16_to_cpu(dip->di_flushiter)) {
                /*
                 * Deal with the wrap case, DI_MAX_FLUSH is less
                 * than smaller numbers
                 */
                if (be16_to_cpu(dip->di_flushiter) == DI_MAX_FLUSH &&
                    dicp->di_flushiter < (DI_MAX_FLUSH >> 1)) {
                        /* do nothing */
                } else {
                        xfs_buf_relse(bp);
                        trace_xfs_log_recover_inode_skip(log, in_f);
                        error = 0;
                        goto error;
                }
        }
        /* Take the opportunity to reset the flush iteration count */
        dicp->di_flushiter = 0;

        if (unlikely(S_ISREG(dicp->di_mode))) {
                if ((dicp->di_format != XFS_DINODE_FMT_EXTENTS) &&
                    (dicp->di_format != XFS_DINODE_FMT_BTREE)) {
                        XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(3)",
                                         XFS_ERRLEVEL_LOW, mp, dicp);
                        xfs_buf_relse(bp);
                        xfs_alert(mp,
                "%s: Bad regular inode log record, rec ptr 0x%p, "
                "ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
                                __func__, item, dip, bp, in_f->ilf_ino);
                        error = EFSCORRUPTED;
                        goto error;
                }
        } else if (unlikely(S_ISDIR(dicp->di_mode))) {
                if ((dicp->di_format != XFS_DINODE_FMT_EXTENTS) &&
                    (dicp->di_format != XFS_DINODE_FMT_BTREE) &&
                    (dicp->di_format != XFS_DINODE_FMT_LOCAL)) {
                        XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(4)",
                                             XFS_ERRLEVEL_LOW, mp, dicp);
                        xfs_buf_relse(bp);
                        xfs_alert(mp,
                "%s: Bad dir inode log record, rec ptr 0x%p, "
                "ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
                                __func__, item, dip, bp, in_f->ilf_ino);
                        error = EFSCORRUPTED;
                        goto error;
                }
        }
        if (unlikely(dicp->di_nextents + dicp->di_anextents > dicp->di_nblocks)){
                XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(5)",
                                     XFS_ERRLEVEL_LOW, mp, dicp);
                xfs_buf_relse(bp);
                xfs_alert(mp,
        "%s: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, "
        "dino bp 0x%p, ino %Ld, total extents = %d, nblocks = %Ld",
                        __func__, item, dip, bp, in_f->ilf_ino,
                        dicp->di_nextents + dicp->di_anextents,
                        dicp->di_nblocks);
                error = EFSCORRUPTED;
                goto error;
        }
        if (unlikely(dicp->di_forkoff > mp->m_sb.sb_inodesize)) {
                XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(6)",
                                     XFS_ERRLEVEL_LOW, mp, dicp);
                xfs_buf_relse(bp);
                xfs_alert(mp,
        "%s: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, "
        "dino bp 0x%p, ino %Ld, forkoff 0x%x", __func__,
                        item, dip, bp, in_f->ilf_ino, dicp->di_forkoff);
                error = EFSCORRUPTED;
                goto error;
        }
        if (unlikely(item->ri_buf[1].i_len > sizeof(struct xfs_icdinode))) {
                XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(7)",
                                     XFS_ERRLEVEL_LOW, mp, dicp);
                xfs_buf_relse(bp);
                xfs_alert(mp,
                        "%s: Bad inode log record length %d, rec ptr 0x%p",
                        __func__, item->ri_buf[1].i_len, item);
                error = EFSCORRUPTED;
                goto error;
        }

        /* The core is in in-core format */
        xfs_dinode_to_disk(dip, item->ri_buf[1].i_addr);

        /* the rest is in on-disk format */
        if (item->ri_buf[1].i_len > sizeof(struct xfs_icdinode)) {
                memcpy((xfs_caddr_t) dip + sizeof(struct xfs_icdinode),
                        item->ri_buf[1].i_addr + sizeof(struct xfs_icdinode),
                        item->ri_buf[1].i_len  - sizeof(struct xfs_icdinode));
        }

        fields = in_f->ilf_fields;
        switch (fields & (XFS_ILOG_DEV | XFS_ILOG_UUID)) {
        case XFS_ILOG_DEV:
                xfs_dinode_put_rdev(dip, in_f->ilf_u.ilfu_rdev);
                break;
        case XFS_ILOG_UUID:
                memcpy(XFS_DFORK_DPTR(dip),
                       &in_f->ilf_u.ilfu_uuid,
                       sizeof(uuid_t));
                break;
        }

        if (in_f->ilf_size == 2)
                goto write_inode_buffer;
        len = item->ri_buf[2].i_len;
        src = item->ri_buf[2].i_addr;
        ASSERT(in_f->ilf_size <= 4);
        ASSERT((in_f->ilf_size == 3) || (fields & XFS_ILOG_AFORK));
        ASSERT(!(fields & XFS_ILOG_DFORK) ||
               (len == in_f->ilf_dsize));

        switch (fields & XFS_ILOG_DFORK) {
        case XFS_ILOG_DDATA:
        case XFS_ILOG_DEXT:
                memcpy(XFS_DFORK_DPTR(dip), src, len);
                break;

        case XFS_ILOG_DBROOT:
                xfs_bmbt_to_bmdr(mp, (struct xfs_btree_block *)src, len,
                                 (xfs_bmdr_block_t *)XFS_DFORK_DPTR(dip),
                                 XFS_DFORK_DSIZE(dip, mp));
                break;

        default:
                /*
                 * There are no data fork flags set.
                 */
                ASSERT((fields & XFS_ILOG_DFORK) == 0);
                break;
        }

        /*
         * If we logged any attribute data, recover it.  There may or
         * may not have been any other non-core data logged in this
         * transaction.
         */
        if (in_f->ilf_fields & XFS_ILOG_AFORK) {
                if (in_f->ilf_fields & XFS_ILOG_DFORK) {
                        attr_index = 3;
                } else {
                        attr_index = 2;
                }
                len = item->ri_buf[attr_index].i_len;
                src = item->ri_buf[attr_index].i_addr;
                ASSERT(len == in_f->ilf_asize);

                switch (in_f->ilf_fields & XFS_ILOG_AFORK) {
                case XFS_ILOG_ADATA:
                case XFS_ILOG_AEXT:
                        dest = XFS_DFORK_APTR(dip);
                        ASSERT(len <= XFS_DFORK_ASIZE(dip, mp));
                        memcpy(dest, src, len);
                        break;

                case XFS_ILOG_ABROOT:
                        dest = XFS_DFORK_APTR(dip);
                        xfs_bmbt_to_bmdr(mp, (struct xfs_btree_block *)src,
                                         len, (xfs_bmdr_block_t*)dest,
                                         XFS_DFORK_ASIZE(dip, mp));
                        break;

                default:
                        xfs_warn(log->l_mp, "%s: Invalid flag", __func__);
                        ASSERT(0);
                        xfs_buf_relse(bp);
                        error = EIO;
                        goto error;
                }
        }

write_inode_buffer:
        ASSERT(bp->b_target->bt_mount == mp);
        bp->b_iodone = xlog_recover_iodone;
        xfs_buf_delwri_queue(bp, buffer_list);
        xfs_buf_relse(bp);
error:
        if (need_free)
                kmem_free(in_f);
        return XFS_ERROR(error);
}

/*
 * Recover QUOTAOFF records. We simply make a note of it in the xlog
 * structure, so that we know not to do any dquot item or dquot buffer recovery,
 * of that type.
 */
STATIC int
xlog_recover_quotaoff_pass1(
        struct xlog                     *log,
        struct xlog_recover_item        *item)
{
        xfs_qoff_logformat_t    *qoff_f = item->ri_buf[0].i_addr;
        ASSERT(qoff_f);

        /*
         * The logitem format's flag tells us if this was user quotaoff,
         * group/project quotaoff or both.
         */
        if (qoff_f->qf_flags & XFS_UQUOTA_ACCT)
                log->l_quotaoffs_flag |= XFS_DQ_USER;
        if (qoff_f->qf_flags & XFS_PQUOTA_ACCT)
                log->l_quotaoffs_flag |= XFS_DQ_PROJ;
        if (qoff_f->qf_flags & XFS_GQUOTA_ACCT)
                log->l_quotaoffs_flag |= XFS_DQ_GROUP;

        return (0);
}

/*
 * Recover a dquot record
 */
STATIC int
xlog_recover_dquot_pass2(
        struct xlog                     *log,
        struct list_head                *buffer_list,
        struct xlog_recover_item        *item)
{
        xfs_mount_t             *mp = log->l_mp;
        xfs_buf_t               *bp;
        struct xfs_disk_dquot   *ddq, *recddq;
        int                     error;
        xfs_dq_logformat_t      *dq_f;
        uint                    type;


        /*
         * Filesystems are required to send in quota flags at mount time.
         */
        if (mp->m_qflags == 0)
                return (0);

        recddq = item->ri_buf[1].i_addr;
        if (recddq == NULL) {
                xfs_alert(log->l_mp, "NULL dquot in %s.", __func__);
                return XFS_ERROR(EIO);
        }
        if (item->ri_buf[1].i_len < sizeof(xfs_disk_dquot_t)) {
                xfs_alert(log->l_mp, "dquot too small (%d) in %s.",
                        item->ri_buf[1].i_len, __func__);
                return XFS_ERROR(EIO);
        }

        /*
         * This type of quotas was turned off, so ignore this record.
         */
        type = recddq->d_flags & (XFS_DQ_USER | XFS_DQ_PROJ | XFS_DQ_GROUP);
        ASSERT(type);
        if (log->l_quotaoffs_flag & type)
                return (0);

        /*
         * At this point we know that quota was _not_ turned off.
         * Since the mount flags are not indicating to us otherwise, this
         * must mean that quota is on, and the dquot needs to be replayed.
         * Remember that we may not have fully recovered the superblock yet,
         * so we can't do the usual trick of looking at the SB quota bits.
         *
         * The other possibility, of course, is that the quota subsystem was
         * removed since the last mount - ENOSYS.
         */
        dq_f = item->ri_buf[0].i_addr;
        ASSERT(dq_f);
        error = xfs_qm_dqcheck(mp, recddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
                           "xlog_recover_dquot_pass2 (log copy)");
        if (error)
                return XFS_ERROR(EIO);
        ASSERT(dq_f->qlf_len == 1);

        error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp, dq_f->qlf_blkno,
                                   XFS_FSB_TO_BB(mp, dq_f->qlf_len), 0, &bp);
        if (error)
                return error;

        ASSERT(bp);
        ddq = (xfs_disk_dquot_t *)xfs_buf_offset(bp, dq_f->qlf_boffset);

        /*
         * At least the magic num portion should be on disk because this
         * was among a chunk of dquots created earlier, and we did some
         * minimal initialization then.
         */
        error = xfs_qm_dqcheck(mp, ddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
                           "xlog_recover_dquot_pass2");
        if (error) {
                xfs_buf_relse(bp);
                return XFS_ERROR(EIO);
        }

        memcpy(ddq, recddq, item->ri_buf[1].i_len);

        ASSERT(dq_f->qlf_size == 2);
        ASSERT(bp->b_target->bt_mount == mp);
        bp->b_iodone = xlog_recover_iodone;
        xfs_buf_delwri_queue(bp, buffer_list);
        xfs_buf_relse(bp);

        return (0);
}

/*
 * This routine is called to create an in-core extent free intent
 * item from the efi format structure which was logged on disk.
 * It allocates an in-core efi, copies the extents from the format
 * structure into it, and adds the efi to the AIL with the given
 * LSN.
 */
STATIC int
xlog_recover_efi_pass2(
        struct xlog                     *log,
        struct xlog_recover_item        *item,
        xfs_lsn_t                       lsn)
{
        int                     error;
        xfs_mount_t             *mp = log->l_mp;
        xfs_efi_log_item_t      *efip;
        xfs_efi_log_format_t    *efi_formatp;

        efi_formatp = item->ri_buf[0].i_addr;

        efip = xfs_efi_init(mp, efi_formatp->efi_nextents);
        if ((error = xfs_efi_copy_format(&(item->ri_buf[0]),
                                         &(efip->efi_format)))) {
                xfs_efi_item_free(efip);
                return error;
        }
        atomic_set(&efip->efi_next_extent, efi_formatp->efi_nextents);

        spin_lock(&log->l_ailp->xa_lock);
        /*
         * xfs_trans_ail_update() drops the AIL lock.
         */
        xfs_trans_ail_update(log->l_ailp, &efip->efi_item, lsn);
        return 0;
}


/*
 * This routine is called when an efd format structure is found in
 * a committed transaction in the log.  It's purpose is to cancel
 * the corresponding efi if it was still in the log.  To do this
 * it searches the AIL for the efi with an id equal to that in the
 * efd format structure.  If we find it, we remove the efi from the
 * AIL and free it.
 */
STATIC int
xlog_recover_efd_pass2(
        struct xlog                     *log,
        struct xlog_recover_item        *item)
{
        xfs_efd_log_format_t    *efd_formatp;
        xfs_efi_log_item_t      *efip = NULL;
        xfs_log_item_t          *lip;
        __uint64_t              efi_id;
        struct xfs_ail_cursor   cur;
        struct xfs_ail          *ailp = log->l_ailp;

        efd_formatp = item->ri_buf[0].i_addr;
        ASSERT((item->ri_buf[0].i_len == (sizeof(xfs_efd_log_format_32_t) +
                ((efd_formatp->efd_nextents - 1) * sizeof(xfs_extent_32_t)))) ||
               (item->ri_buf[0].i_len == (sizeof(xfs_efd_log_format_64_t) +
                ((efd_formatp->efd_nextents - 1) * sizeof(xfs_extent_64_t)))));
        efi_id = efd_formatp->efd_efi_id;

        /*
         * Search for the efi with the id in the efd format structure
         * in the AIL.
         */
        spin_lock(&ailp->xa_lock);
        lip = xfs_trans_ail_cursor_first(ailp, &cur, 0);
        while (lip != NULL) {
                if (lip->li_type == XFS_LI_EFI) {
                        efip = (xfs_efi_log_item_t *)lip;
                        if (efip->efi_format.efi_id == efi_id) {
                                /*
                                 * xfs_trans_ail_delete() drops the
                                 * AIL lock.
                                 */
                                xfs_trans_ail_delete(ailp, lip,
                                                     SHUTDOWN_CORRUPT_INCORE);
                                xfs_efi_item_free(efip);
                                spin_lock(&ailp->xa_lock);
                                break;
                        }
                }
                lip = xfs_trans_ail_cursor_next(ailp, &cur);
        }
        xfs_trans_ail_cursor_done(ailp, &cur);
        spin_unlock(&ailp->xa_lock);

        return 0;
}

/*
 * Free up any resources allocated by the transaction
 *
 * Remember that EFIs, EFDs, and IUNLINKs are handled later.
 */
STATIC void
xlog_recover_free_trans(
        struct xlog_recover     *trans)
{
        xlog_recover_item_t     *item, *n;
        int                     i;

        list_for_each_entry_safe(item, n, &trans->r_itemq, ri_list) {
                /* Free the regions in the item. */
                list_del(&item->ri_list);
                for (i = 0; i < item->ri_cnt; i++)
                        kmem_free(item->ri_buf[i].i_addr);
                /* Free the item itself */
                kmem_free(item->ri_buf);
                kmem_free(item);
        }
        /* Free the transaction recover structure */
        kmem_free(trans);
}

STATIC int
xlog_recover_commit_pass1(
        struct xlog                     *log,
        struct xlog_recover             *trans,
        struct xlog_recover_item        *item)
{
        trace_xfs_log_recover_item_recover(log, trans, item, XLOG_RECOVER_PASS1);

        switch (ITEM_TYPE(item)) {
        case XFS_LI_BUF:
                return xlog_recover_buffer_pass1(log, item);
        case XFS_LI_QUOTAOFF:
                return xlog_recover_quotaoff_pass1(log, item);
        case XFS_LI_INODE:
        case XFS_LI_EFI:
        case XFS_LI_EFD:
        case XFS_LI_DQUOT:
                /* nothing to do in pass 1 */
                return 0;
        default:
                xfs_warn(log->l_mp, "%s: invalid item type (%d)",
                        __func__, ITEM_TYPE(item));
                ASSERT(0);
                return XFS_ERROR(EIO);
        }
}

STATIC int
xlog_recover_commit_pass2(
        struct xlog                     *log,
        struct xlog_recover             *trans,
        struct list_head                *buffer_list,
        struct xlog_recover_item        *item)
{
        trace_xfs_log_recover_item_recover(log, trans, item, XLOG_RECOVER_PASS2);

        switch (ITEM_TYPE(item)) {
        case XFS_LI_BUF:
                return xlog_recover_buffer_pass2(log, buffer_list, item);
        case XFS_LI_INODE:
                return xlog_recover_inode_pass2(log, buffer_list, item);
        case XFS_LI_EFI:
                return xlog_recover_efi_pass2(log, item, trans->r_lsn);
        case XFS_LI_EFD:
                return xlog_recover_efd_pass2(log, item);
        case XFS_LI_DQUOT:
                return xlog_recover_dquot_pass2(log, buffer_list, item);
        case XFS_LI_QUOTAOFF:
                /* nothing to do in pass2 */
                return 0;
        default:
                xfs_warn(log->l_mp, "%s: invalid item type (%d)",
                        __func__, ITEM_TYPE(item));
                ASSERT(0);
                return XFS_ERROR(EIO);
        }
}

/*
 * Perform the transaction.
 *
 * If the transaction modifies a buffer or inode, do it now.  Otherwise,
 * EFIs and EFDs get queued up by adding entries into the AIL for them.
 */
STATIC int
xlog_recover_commit_trans(
        struct xlog             *log,
        struct xlog_recover     *trans,
        int                     pass)
{
        int                     error = 0, error2;
        xlog_recover_item_t     *item;
        LIST_HEAD               (buffer_list);

        hlist_del(&trans->r_list);

        error = xlog_recover_reorder_trans(log, trans, pass);
        if (error)
                return error;

        list_for_each_entry(item, &trans->r_itemq, ri_list) {
                switch (pass) {
                case XLOG_RECOVER_PASS1:
                        error = xlog_recover_commit_pass1(log, trans, item);
                        break;
                case XLOG_RECOVER_PASS2:
                        error = xlog_recover_commit_pass2(log, trans,
                                                          &buffer_list, item);
                        break;
                default:
                        ASSERT(0);
                }

                if (error)
                        goto out;
        }

        xlog_recover_free_trans(trans);

out:
        error2 = xfs_buf_delwri_submit(&buffer_list);
        return error ? error : error2;
}

STATIC int
xlog_recover_unmount_trans(
        struct xlog             *log,
        struct xlog_recover     *trans)
{
        /* Do nothing now */
        xfs_warn(log->l_mp, "%s: Unmount LR", __func__);
        return 0;
}

/*
 * There are two valid states of the r_state field.  0 indicates that the
 * transaction structure is in a normal state.  We have either seen the
 * start of the transaction or the last operation we added was not a partial
 * operation.  If the last operation we added to the transaction was a
 * partial operation, we need to mark r_state with XLOG_WAS_CONT_TRANS.
 *
 * NOTE: skip LRs with 0 data length.
 */
STATIC int
xlog_recover_process_data(
        struct xlog             *log,
        struct hlist_head       rhash[],
        struct xlog_rec_header  *rhead,
        xfs_caddr_t             dp,
        int                     pass)
{
        xfs_caddr_t             lp;
        int                     num_logops;
        xlog_op_header_t        *ohead;
        xlog_recover_t          *trans;
        xlog_tid_t              tid;
        int                     error;
        unsigned long           hash;
        uint                    flags;

        lp = dp + be32_to_cpu(rhead->h_len);
        num_logops = be32_to_cpu(rhead->h_num_logops);

        /* check the log format matches our own - else we can't recover */
        if (xlog_header_check_recover(log->l_mp, rhead))
                return (XFS_ERROR(EIO));

        while ((dp < lp) && num_logops) {
                ASSERT(dp + sizeof(xlog_op_header_t) <= lp);
                ohead = (xlog_op_header_t *)dp;
                dp += sizeof(xlog_op_header_t);
                if (ohead->oh_clientid != XFS_TRANSACTION &&
                    ohead->oh_clientid != XFS_LOG) {
                        xfs_warn(log->l_mp, "%s: bad clientid 0x%x",
                                        __func__, ohead->oh_clientid);
                        ASSERT(0);
                        return (XFS_ERROR(EIO));
                }
                tid = be32_to_cpu(ohead->oh_tid);
                hash = XLOG_RHASH(tid);
                trans = xlog_recover_find_tid(&rhash[hash], tid);
                if (trans == NULL) {               /* not found; add new tid */
                        if (ohead->oh_flags & XLOG_START_TRANS)
                                xlog_recover_new_tid(&rhash[hash], tid,
                                        be64_to_cpu(rhead->h_lsn));
                } else {
                        if (dp + be32_to_cpu(ohead->oh_len) > lp) {
                                xfs_warn(log->l_mp, "%s: bad length 0x%x",
                                        __func__, be32_to_cpu(ohead->oh_len));
                                WARN_ON(1);
                                return (XFS_ERROR(EIO));
                        }
                        flags = ohead->oh_flags & ~XLOG_END_TRANS;
                        if (flags & XLOG_WAS_CONT_TRANS)
                                flags &= ~XLOG_CONTINUE_TRANS;
                        switch (flags) {
                        case XLOG_COMMIT_TRANS:
                                error = xlog_recover_commit_trans(log,
                                                                trans, pass);
                                break;
                        case XLOG_UNMOUNT_TRANS:
                                error = xlog_recover_unmount_trans(log, trans);
                                break;
                        case XLOG_WAS_CONT_TRANS:
                                error = xlog_recover_add_to_cont_trans(log,
                                                trans, dp,
                                                be32_to_cpu(ohead->oh_len));
                                break;
                        case XLOG_START_TRANS:
                                xfs_warn(log->l_mp, "%s: bad transaction",
                                        __func__);
                                ASSERT(0);
                                error = XFS_ERROR(EIO);
                                break;
                        case 0:
                        case XLOG_CONTINUE_TRANS:
                                error = xlog_recover_add_to_trans(log, trans,
                                                dp, be32_to_cpu(ohead->oh_len));
                                break;
                        default:
                                xfs_warn(log->l_mp, "%s: bad flag 0x%x",
                                        __func__, flags);
                                ASSERT(0);
                                error = XFS_ERROR(EIO);
                                break;
                        }
                        if (error)
                                return error;
                }
                dp += be32_to_cpu(ohead->oh_len);
                num_logops--;
        }
        return 0;
}

/*
 * Process an extent free intent item that was recovered from
 * the log.  We need to free the extents that it describes.
 */
STATIC int
xlog_recover_process_efi(
        xfs_mount_t             *mp,
        xfs_efi_log_item_t      *efip)
{
        xfs_efd_log_item_t      *efdp;
        xfs_trans_t             *tp;
        int                     i;
        int                     error = 0;
        xfs_extent_t            *extp;
        xfs_fsblock_t           startblock_fsb;

        ASSERT(!test_bit(XFS_EFI_RECOVERED, &efip->efi_flags));

        /*
         * First check the validity of the extents described by the
         * EFI.  If any are bad, then assume that all are bad and
         * just toss the EFI.
         */
        for (i = 0; i < efip->efi_format.efi_nextents; i++) {
                extp = &(efip->efi_format.efi_extents[i]);
                startblock_fsb = XFS_BB_TO_FSB(mp,
                                   XFS_FSB_TO_DADDR(mp, extp->ext_start));
                if ((startblock_fsb == 0) ||
                    (extp->ext_len == 0) ||
                    (startblock_fsb >= mp->m_sb.sb_dblocks) ||
                    (extp->ext_len >= mp->m_sb.sb_agblocks)) {
                        /*
                         * This will pull the EFI from the AIL and
                         * free the memory associated with it.
                         */
                        xfs_efi_release(efip, efip->efi_format.efi_nextents);
                        return XFS_ERROR(EIO);
                }
        }

        tp = xfs_trans_alloc(mp, 0);
        error = xfs_trans_reserve(tp, 0, XFS_ITRUNCATE_LOG_RES(mp), 0, 0, 0);
        if (error)
                goto abort_error;
        efdp = xfs_trans_get_efd(tp, efip, efip->efi_format.efi_nextents);

        for (i = 0; i < efip->efi_format.efi_nextents; i++) {
                extp = &(efip->efi_format.efi_extents[i]);
                error = xfs_free_extent(tp, extp->ext_start, extp->ext_len);
                if (error)
                        goto abort_error;
                xfs_trans_log_efd_extent(tp, efdp, extp->ext_start,
                                         extp->ext_len);
        }

        set_bit(XFS_EFI_RECOVERED, &efip->efi_flags);
        error = xfs_trans_commit(tp, 0);
        return error;

abort_error:
        xfs_trans_cancel(tp, XFS_TRANS_ABORT);
        return error;
}

/*
 * When this is called, all of the EFIs which did not have
 * corresponding EFDs should be in the AIL.  What we do now
 * is free the extents associated with each one.
 *
 * Since we process the EFIs in normal transactions, they
 * will be removed at some point after the commit.  This prevents
 * us from just walking down the list processing each one.
 * We'll use a flag in the EFI to skip those that we've already
 * processed and use the AIL iteration mechanism's generation
 * count to try to speed this up at least a bit.
 *
 * When we start, we know that the EFIs are the only things in
 * the AIL.  As we process them, however, other items are added
 * to the AIL.  Since everything added to the AIL must come after
 * everything already in the AIL, we stop processing as soon as
 * we see something other than an EFI in the AIL.
 */
STATIC int
xlog_recover_process_efis(
        struct xlog     *log)
{
        xfs_log_item_t          *lip;
        xfs_efi_log_item_t      *efip;