xfs: split in-core and on-disk inode log item fields
authorChristoph Hellwig <hch@infradead.org>
Wed, 29 Feb 2012 09:53:54 +0000 (09:53 +0000)
committerBen Myers <bpm@sgi.com>
Tue, 13 Mar 2012 22:08:17 +0000 (17:08 -0500)
Add a new ili_fields member to the inode log item to isolate the in-memory
flags from the ones that actually go to the log.  This will allow tracking
timestamp-only updates for fdatasync and O_DSYNC in the next patch and
prepares for divorcing the on-disk log format from the in-memory log item
a little further down the road.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
fs/xfs/xfs_dfrag.c
fs/xfs/xfs_inode.c
fs/xfs/xfs_inode_item.c
fs/xfs/xfs_inode_item.h
fs/xfs/xfs_trans_inode.c

index dd974a5..1137bbc 100644 (file)
@@ -215,7 +215,7 @@ xfs_swap_extents(
        xfs_trans_t     *tp;
        xfs_bstat_t     *sbp = &sxp->sx_stat;
        xfs_ifork_t     *tempifp, *ifp, *tifp;
-       int             ilf_fields, tilf_fields;
+       int             src_log_flags, target_log_flags;
        int             error = 0;
        int             aforkblks = 0;
        int             taforkblks = 0;
@@ -385,9 +385,8 @@ xfs_swap_extents(
        tip->i_delayed_blks = ip->i_delayed_blks;
        ip->i_delayed_blks = 0;
 
-       ilf_fields = XFS_ILOG_CORE;
-
-       switch(ip->i_d.di_format) {
+       src_log_flags = XFS_ILOG_CORE;
+       switch (ip->i_d.di_format) {
        case XFS_DINODE_FMT_EXTENTS:
                /* If the extents fit in the inode, fix the
                 * pointer.  Otherwise it's already NULL or
@@ -397,16 +396,15 @@ xfs_swap_extents(
                        ifp->if_u1.if_extents =
                                ifp->if_u2.if_inline_ext;
                }
-               ilf_fields |= XFS_ILOG_DEXT;
+               src_log_flags |= XFS_ILOG_DEXT;
                break;
        case XFS_DINODE_FMT_BTREE:
-               ilf_fields |= XFS_ILOG_DBROOT;
+               src_log_flags |= XFS_ILOG_DBROOT;
                break;
        }
 
-       tilf_fields = XFS_ILOG_CORE;
-
-       switch(tip->i_d.di_format) {
+       target_log_flags = XFS_ILOG_CORE;
+       switch (tip->i_d.di_format) {
        case XFS_DINODE_FMT_EXTENTS:
                /* If the extents fit in the inode, fix the
                 * pointer.  Otherwise it's already NULL or
@@ -416,10 +414,10 @@ xfs_swap_extents(
                        tifp->if_u1.if_extents =
                                tifp->if_u2.if_inline_ext;
                }
-               tilf_fields |= XFS_ILOG_DEXT;
+               target_log_flags |= XFS_ILOG_DEXT;
                break;
        case XFS_DINODE_FMT_BTREE:
-               tilf_fields |= XFS_ILOG_DBROOT;
+               target_log_flags |= XFS_ILOG_DBROOT;
                break;
        }
 
@@ -427,8 +425,8 @@ xfs_swap_extents(
        xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
        xfs_trans_ijoin(tp, tip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 
-       xfs_trans_log_inode(tp, ip,  ilf_fields);
-       xfs_trans_log_inode(tp, tip, tilf_fields);
+       xfs_trans_log_inode(tp, ip,  src_log_flags);
+       xfs_trans_log_inode(tp, tip, target_log_flags);
 
        /*
         * If this is a synchronous mount, make sure that the
index 7ce9ccb..bc46c0a 100644 (file)
@@ -1661,8 +1661,8 @@ retry:
                                continue;
                        }
 
-                       iip->ili_last_fields = iip->ili_format.ilf_fields;
-                       iip->ili_format.ilf_fields = 0;
+                       iip->ili_last_fields = iip->ili_fields;
+                       iip->ili_fields = 0;
                        iip->ili_logged = 1;
                        xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
                                                &iip->ili_item.li_lsn);
@@ -2176,7 +2176,7 @@ xfs_iflush_fork(
        mp = ip->i_mount;
        switch (XFS_IFORK_FORMAT(ip, whichfork)) {
        case XFS_DINODE_FMT_LOCAL:
-               if ((iip->ili_format.ilf_fields & dataflag[whichfork]) &&
+               if ((iip->ili_fields & dataflag[whichfork]) &&
                    (ifp->if_bytes > 0)) {
                        ASSERT(ifp->if_u1.if_data != NULL);
                        ASSERT(ifp->if_bytes <= XFS_IFORK_SIZE(ip, whichfork));
@@ -2186,8 +2186,8 @@ xfs_iflush_fork(
 
        case XFS_DINODE_FMT_EXTENTS:
                ASSERT((ifp->if_flags & XFS_IFEXTENTS) ||
-                      !(iip->ili_format.ilf_fields & extflag[whichfork]));
-               if ((iip->ili_format.ilf_fields & extflag[whichfork]) &&
+                      !(iip->ili_fields & extflag[whichfork]));
+               if ((iip->ili_fields & extflag[whichfork]) &&
                    (ifp->if_bytes > 0)) {
                        ASSERT(xfs_iext_get_ext(ifp, 0));
                        ASSERT(XFS_IFORK_NEXTENTS(ip, whichfork) > 0);
@@ -2197,7 +2197,7 @@ xfs_iflush_fork(
                break;
 
        case XFS_DINODE_FMT_BTREE:
-               if ((iip->ili_format.ilf_fields & brootflag[whichfork]) &&
+               if ((iip->ili_fields & brootflag[whichfork]) &&
                    (ifp->if_broot_bytes > 0)) {
                        ASSERT(ifp->if_broot != NULL);
                        ASSERT(ifp->if_broot_bytes <=
@@ -2210,14 +2210,14 @@ xfs_iflush_fork(
                break;
 
        case XFS_DINODE_FMT_DEV:
-               if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
+               if (iip->ili_fields & XFS_ILOG_DEV) {
                        ASSERT(whichfork == XFS_DATA_FORK);
                        xfs_dinode_put_rdev(dip, ip->i_df.if_u2.if_rdev);
                }
                break;
 
        case XFS_DINODE_FMT_UUID:
-               if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
+               if (iip->ili_fields & XFS_ILOG_UUID) {
                        ASSERT(whichfork == XFS_DATA_FORK);
                        memcpy(XFS_DFORK_DPTR(dip),
                               &ip->i_df.if_u2.if_uuid,
@@ -2451,7 +2451,7 @@ xfs_iflush(
         */
        if (XFS_FORCED_SHUTDOWN(mp)) {
                if (iip)
-                       iip->ili_format.ilf_fields = 0;
+                       iip->ili_fields = 0;
                xfs_ifunlock(ip);
                return XFS_ERROR(EIO);
        }
@@ -2641,36 +2641,33 @@ xfs_iflush_int(
        xfs_inobp_check(mp, bp);
 
        /*
-        * We've recorded everything logged in the inode, so we'd
-        * like to clear the ilf_fields bits so we don't log and
-        * flush things unnecessarily.  However, we can't stop
-        * logging all this information until the data we've copied
-        * into the disk buffer is written to disk.  If we did we might
-        * overwrite the copy of the inode in the log with all the
-        * data after re-logging only part of it, and in the face of
-        * a crash we wouldn't have all the data we need to recover.
+        * We've recorded everything logged in the inode, so we'd like to clear
+        * the ili_fields bits so we don't log and flush things unnecessarily.
+        * However, we can't stop logging all this information until the data
+        * we've copied into the disk buffer is written to disk.  If we did we
+        * might overwrite the copy of the inode in the log with all the data
+        * after re-logging only part of it, and in the face of a crash we
+        * wouldn't have all the data we need to recover.
         *
-        * What we do is move the bits to the ili_last_fields field.
-        * When logging the inode, these bits are moved back to the
-        * ilf_fields field.  In the xfs_iflush_done() routine we
-        * clear ili_last_fields, since we know that the information
-        * those bits represent is permanently on disk.  As long as
-        * the flush completes before the inode is logged again, then
-        * both ilf_fields and ili_last_fields will be cleared.
+        * What we do is move the bits to the ili_last_fields field.  When
+        * logging the inode, these bits are moved back to the ili_fields field.
+        * In the xfs_iflush_done() routine we clear ili_last_fields, since we
+        * know that the information those bits represent is permanently on
+        * disk.  As long as the flush completes before the inode is logged
+        * again, then both ili_fields and ili_last_fields will be cleared.
         *
-        * We can play with the ilf_fields bits here, because the inode
-        * lock must be held exclusively in order to set bits there
-        * and the flush lock protects the ili_last_fields bits.
-        * Set ili_logged so the flush done
-        * routine can tell whether or not to look in the AIL.
-        * Also, store the current LSN of the inode so that we can tell
-        * whether the item has moved in the AIL from xfs_iflush_done().
-        * In order to read the lsn we need the AIL lock, because
-        * it is a 64 bit value that cannot be read atomically.
+        * We can play with the ili_fields bits here, because the inode lock
+        * must be held exclusively in order to set bits there and the flush
+        * lock protects the ili_last_fields bits.  Set ili_logged so the flush
+        * done routine can tell whether or not to look in the AIL.  Also, store
+        * the current LSN of the inode so that we can tell whether the item has
+        * moved in the AIL from xfs_iflush_done().  In order to read the lsn we
+        * need the AIL lock, because it is a 64 bit value that cannot be read
+        * atomically.
         */
-       if (iip != NULL && iip->ili_format.ilf_fields != 0) {
-               iip->ili_last_fields = iip->ili_format.ilf_fields;
-               iip->ili_format.ilf_fields = 0;
+       if (iip != NULL && iip->ili_fields != 0) {
+               iip->ili_last_fields = iip->ili_fields;
+               iip->ili_fields = 0;
                iip->ili_logged = 1;
 
                xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
index 965d3d0..8becef4 100644 (file)
@@ -59,20 +59,20 @@ xfs_inode_item_size(
 
        switch (ip->i_d.di_format) {
        case XFS_DINODE_FMT_EXTENTS:
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_DEXT) &&
+               if ((iip->ili_fields & XFS_ILOG_DEXT) &&
                    ip->i_d.di_nextents > 0 &&
                    ip->i_df.if_bytes > 0)
                        nvecs++;
                break;
 
        case XFS_DINODE_FMT_BTREE:
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) &&
+               if ((iip->ili_fields & XFS_ILOG_DBROOT) &&
                    ip->i_df.if_broot_bytes > 0)
                        nvecs++;
                break;
 
        case XFS_DINODE_FMT_LOCAL:
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_DDATA) &&
+               if ((iip->ili_fields & XFS_ILOG_DDATA) &&
                    ip->i_df.if_bytes > 0)
                        nvecs++;
                break;
@@ -95,20 +95,20 @@ xfs_inode_item_size(
         */
        switch (ip->i_d.di_aformat) {
        case XFS_DINODE_FMT_EXTENTS:
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_AEXT) &&
+               if ((iip->ili_fields & XFS_ILOG_AEXT) &&
                    ip->i_d.di_anextents > 0 &&
                    ip->i_afp->if_bytes > 0)
                        nvecs++;
                break;
 
        case XFS_DINODE_FMT_BTREE:
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) &&
+               if ((iip->ili_fields & XFS_ILOG_ABROOT) &&
                    ip->i_afp->if_broot_bytes > 0)
                        nvecs++;
                break;
 
        case XFS_DINODE_FMT_LOCAL:
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_ADATA) &&
+               if ((iip->ili_fields & XFS_ILOG_ADATA) &&
                    ip->i_afp->if_bytes > 0)
                        nvecs++;
                break;
@@ -185,7 +185,6 @@ xfs_inode_item_format(
        vecp->i_type = XLOG_REG_TYPE_ICORE;
        vecp++;
        nvecs++;
-       iip->ili_format.ilf_fields |= XFS_ILOG_CORE;
 
        /*
         * If this is really an old format inode, then we need to
@@ -218,11 +217,11 @@ xfs_inode_item_format(
 
        switch (ip->i_d.di_format) {
        case XFS_DINODE_FMT_EXTENTS:
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
                          XFS_ILOG_DEV | XFS_ILOG_UUID);
 
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_DEXT) &&
+               if ((iip->ili_fields & XFS_ILOG_DEXT) &&
                    ip->i_d.di_nextents > 0 &&
                    ip->i_df.if_bytes > 0) {
                        ASSERT(ip->i_df.if_u1.if_extents != NULL);
@@ -251,16 +250,16 @@ xfs_inode_item_format(
                        vecp++;
                        nvecs++;
                } else {
-                       iip->ili_format.ilf_fields &= ~XFS_ILOG_DEXT;
+                       iip->ili_fields &= ~XFS_ILOG_DEXT;
                }
                break;
 
        case XFS_DINODE_FMT_BTREE:
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_DDATA | XFS_ILOG_DEXT |
                          XFS_ILOG_DEV | XFS_ILOG_UUID);
 
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) &&
+               if ((iip->ili_fields & XFS_ILOG_DBROOT) &&
                    ip->i_df.if_broot_bytes > 0) {
                        ASSERT(ip->i_df.if_broot != NULL);
                        vecp->i_addr = ip->i_df.if_broot;
@@ -270,7 +269,7 @@ xfs_inode_item_format(
                        nvecs++;
                        iip->ili_format.ilf_dsize = ip->i_df.if_broot_bytes;
                } else {
-                       ASSERT(!(iip->ili_format.ilf_fields &
+                       ASSERT(!(iip->ili_fields &
                                 XFS_ILOG_DBROOT));
 #ifdef XFS_TRANS_DEBUG
                        if (iip->ili_root_size > 0) {
@@ -283,15 +282,15 @@ xfs_inode_item_format(
                                ASSERT(ip->i_df.if_broot_bytes == 0);
                        }
 #endif
-                       iip->ili_format.ilf_fields &= ~XFS_ILOG_DBROOT;
+                       iip->ili_fields &= ~XFS_ILOG_DBROOT;
                }
                break;
 
        case XFS_DINODE_FMT_LOCAL:
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_DEXT | XFS_ILOG_DBROOT |
                          XFS_ILOG_DEV | XFS_ILOG_UUID);
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_DDATA) &&
+               if ((iip->ili_fields & XFS_ILOG_DDATA) &&
                    ip->i_df.if_bytes > 0) {
                        ASSERT(ip->i_df.if_u1.if_data != NULL);
                        ASSERT(ip->i_d.di_size > 0);
@@ -311,25 +310,25 @@ xfs_inode_item_format(
                        nvecs++;
                        iip->ili_format.ilf_dsize = (unsigned)data_bytes;
                } else {
-                       iip->ili_format.ilf_fields &= ~XFS_ILOG_DDATA;
+                       iip->ili_fields &= ~XFS_ILOG_DDATA;
                }
                break;
 
        case XFS_DINODE_FMT_DEV:
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
                          XFS_ILOG_DEXT | XFS_ILOG_UUID);
-               if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
+               if (iip->ili_fields & XFS_ILOG_DEV) {
                        iip->ili_format.ilf_u.ilfu_rdev =
                                ip->i_df.if_u2.if_rdev;
                }
                break;
 
        case XFS_DINODE_FMT_UUID:
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
                          XFS_ILOG_DEXT | XFS_ILOG_DEV);
-               if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
+               if (iip->ili_fields & XFS_ILOG_UUID) {
                        iip->ili_format.ilf_u.ilfu_uuid =
                                ip->i_df.if_u2.if_uuid;
                }
@@ -344,18 +343,17 @@ xfs_inode_item_format(
         * If there are no attributes associated with the file, then we're done.
         */
        if (!XFS_IFORK_Q(ip)) {
-               iip->ili_format.ilf_size = nvecs;
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT);
-               return;
+               goto out;
        }
 
        switch (ip->i_d.di_aformat) {
        case XFS_DINODE_FMT_EXTENTS:
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT);
 
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_AEXT) &&
+               if ((iip->ili_fields & XFS_ILOG_AEXT) &&
                    ip->i_d.di_anextents > 0 &&
                    ip->i_afp->if_bytes > 0) {
                        ASSERT(ip->i_afp->if_bytes / sizeof(xfs_bmbt_rec_t) ==
@@ -378,15 +376,15 @@ xfs_inode_item_format(
                        vecp++;
                        nvecs++;
                } else {
-                       iip->ili_format.ilf_fields &= ~XFS_ILOG_AEXT;
+                       iip->ili_fields &= ~XFS_ILOG_AEXT;
                }
                break;
 
        case XFS_DINODE_FMT_BTREE:
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_ADATA | XFS_ILOG_AEXT);
 
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) &&
+               if ((iip->ili_fields & XFS_ILOG_ABROOT) &&
                    ip->i_afp->if_broot_bytes > 0) {
                        ASSERT(ip->i_afp->if_broot != NULL);
 
@@ -397,15 +395,15 @@ xfs_inode_item_format(
                        nvecs++;
                        iip->ili_format.ilf_asize = ip->i_afp->if_broot_bytes;
                } else {
-                       iip->ili_format.ilf_fields &= ~XFS_ILOG_ABROOT;
+                       iip->ili_fields &= ~XFS_ILOG_ABROOT;
                }
                break;
 
        case XFS_DINODE_FMT_LOCAL:
-               iip->ili_format.ilf_fields &=
+               iip->ili_fields &=
                        ~(XFS_ILOG_AEXT | XFS_ILOG_ABROOT);
 
-               if ((iip->ili_format.ilf_fields & XFS_ILOG_ADATA) &&
+               if ((iip->ili_fields & XFS_ILOG_ADATA) &&
                    ip->i_afp->if_bytes > 0) {
                        ASSERT(ip->i_afp->if_u1.if_data != NULL);
 
@@ -424,7 +422,7 @@ xfs_inode_item_format(
                        nvecs++;
                        iip->ili_format.ilf_asize = (unsigned)data_bytes;
                } else {
-                       iip->ili_format.ilf_fields &= ~XFS_ILOG_ADATA;
+                       iip->ili_fields &= ~XFS_ILOG_ADATA;
                }
                break;
 
@@ -433,6 +431,14 @@ xfs_inode_item_format(
                break;
        }
 
+out:
+       /*
+        * Now update the log format that goes out to disk from the in-core
+        * values.  We always write the inode core to make the arithmetic
+        * games in recovery easier, which isn't a big deal as just about any
+        * transaction would dirty it anyway.
+        */
+       iip->ili_format.ilf_fields = XFS_ILOG_CORE | iip->ili_fields;
        iip->ili_format.ilf_size = nvecs;
 }
 
@@ -517,7 +523,7 @@ xfs_inode_item_trylock(
 
 #ifdef DEBUG
        if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
-               ASSERT(iip->ili_format.ilf_fields != 0);
+               ASSERT(iip->ili_fields != 0);
                ASSERT(iip->ili_logged == 0);
                ASSERT(lip->li_flags & XFS_LI_IN_AIL);
        }
@@ -549,7 +555,7 @@ xfs_inode_item_unlock(
        if (iip->ili_extents_buf != NULL) {
                ASSERT(ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS);
                ASSERT(ip->i_d.di_nextents > 0);
-               ASSERT(iip->ili_format.ilf_fields & XFS_ILOG_DEXT);
+               ASSERT(iip->ili_fields & XFS_ILOG_DEXT);
                ASSERT(ip->i_df.if_bytes > 0);
                kmem_free(iip->ili_extents_buf);
                iip->ili_extents_buf = NULL;
@@ -557,7 +563,7 @@ xfs_inode_item_unlock(
        if (iip->ili_aextents_buf != NULL) {
                ASSERT(ip->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS);
                ASSERT(ip->i_d.di_anextents > 0);
-               ASSERT(iip->ili_format.ilf_fields & XFS_ILOG_AEXT);
+               ASSERT(iip->ili_fields & XFS_ILOG_AEXT);
                ASSERT(ip->i_afp->if_bytes > 0);
                kmem_free(iip->ili_aextents_buf);
                iip->ili_aextents_buf = NULL;
@@ -672,8 +678,7 @@ xfs_inode_item_push(
         * lock without sleeping, then there must not have been
         * anyone in the process of flushing the inode.
         */
-       ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) ||
-              iip->ili_format.ilf_fields != 0);
+       ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) || iip->ili_fields != 0);
 
        /*
         * Push the inode to it's backing buffer. This will not remove the
@@ -896,7 +901,7 @@ xfs_iflush_abort(
                 * Clear the inode logging fields so no more flushes are
                 * attempted.
                 */
-               iip->ili_format.ilf_fields = 0;
+               iip->ili_fields = 0;
        }
        /*
         * Release the inode's flush lock since we're done with it.
index 25784b0..bc183d8 100644 (file)
@@ -134,6 +134,7 @@ typedef struct xfs_inode_log_item {
        unsigned short          ili_lock_flags;    /* lock flags */
        unsigned short          ili_logged;        /* flushed logged data */
        unsigned int            ili_last_fields;   /* fields when flushed */
+       unsigned int            ili_fields;        /* fields to be logged */
        struct xfs_bmbt_rec     *ili_extents_buf;  /* array of logged
                                                      data exts */
        struct xfs_bmbt_rec     *ili_aextents_buf; /* array of logged
@@ -148,8 +149,7 @@ typedef struct xfs_inode_log_item {
 
 static inline int xfs_inode_clean(xfs_inode_t *ip)
 {
-       return !ip->i_itemp ||
-               !(ip->i_itemp->ili_format.ilf_fields & XFS_ILOG_ALL);
+       return !ip->i_itemp || !(ip->i_itemp->ili_fields & XFS_ILOG_ALL);
 }
 
 extern void xfs_inode_item_init(struct xfs_inode *, struct xfs_mount *);
index 892763e..7a7442c 100644 (file)
@@ -130,12 +130,12 @@ xfs_trans_log_inode(
        /*
         * Always OR in the bits from the ili_last_fields field.
         * This is to coordinate with the xfs_iflush() and xfs_iflush_done()
-        * routines in the eventual clearing of the ilf_fields bits.
+        * routines in the eventual clearing of the ili_fields bits.
         * See the big comment in xfs_iflush() for an explanation of
         * this coordination mechanism.
         */
        flags |= ip->i_itemp->ili_last_fields;
-       ip->i_itemp->ili_format.ilf_fields |= flags;
+       ip->i_itemp->ili_fields |= flags;
 }
 
 #ifdef XFS_TRANS_DEBUG