4 files changed, 29 insertions, 153 deletions

diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index 91a54a79a09b..ed24435af651 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -316,11 +316,21 @@ xfs_sync_fsdata(
 }
 
 /*
- * First stage of freeze - no more writers will make progress now we are here,
+ * When remounting a filesystem read-only or freezing the filesystem, we have
+ * two phases to execute. This first phase is syncing the data before we
+ * quiesce the filesystem, and the second is flushing all the inodes out after
+ * we've waited for all the transactions created by the first phase to
+ * complete. The second phase ensures that the inodes are written to their
+ * location on disk rather than just existing in transactions in the log. This
+ * means after a quiesce there is no log replay required to write the inodes to
+ * disk (this is the main difference between a sync and a quiesce).
+ */
+/*
+ * First stage of freeze - no writers will make progress now we are here,
  * so we flush delwri and delalloc buffers here, then wait for all I/O to
  * complete.  Data is frozen at that point. Metadata is not frozen,
- * transactions can still occur here so don't bother flushing the buftarg (i.e
- * SYNC_QUIESCE) because it'll just get dirty again.
+ * transactions can still occur here so don't bother flushing the buftarg
+ * because it'll just get dirty again.
  */
 int
 xfs_quiesce_data(
@@ -337,11 +347,10 @@ xfs_quiesce_data(
         xfs_sync_inodes(mp, SYNC_DELWRI|SYNC_WAIT|SYNC_IOWAIT);
         XFS_QM_DQSYNC(mp, SYNC_WAIT);
 
-        /* write superblock and hoover shutdown errors */
+        /* write superblock and hoover up shutdown errors */
         error = xfs_sync_fsdata(mp, 0);
 
-        /* flush devices */
-        XFS_bflush(mp->m_ddev_targp);
+        /* flush data-only devices */
         if (mp->m_rtdev_targp)
                 XFS_bflush(mp->m_rtdev_targp);
 
@@ -349,117 +358,6 @@ xfs_quiesce_data(
 }
 
 /*
- * xfs_sync flushes any pending I/O to file system vfsp.
- *
- * This routine is called by vfs_sync() to make sure that things make it
- * out to disk eventually, on sync() system calls to flush out everything,
- * and when the file system is unmounted.  For the vfs_sync() case, all
- * we really need to do is sync out the log to make all of our meta-data
- * updates permanent (except for timestamps).  For calls from pflushd(),
- * dirty pages are kept moving by calling pdflush() on the inodes
- * containing them.  We also flush the inodes that we can lock without
- * sleeping and the superblock if we can lock it without sleeping from
- * vfs_sync() so that items at the tail of the log are always moving out.
- *
- * Flags:
- *      SYNC_BDFLUSH - We're being called from vfs_sync() so we don't want
- *                     to sleep if we can help it.  All we really need
- *                     to do is ensure that the log is synced at least
- *                     periodically.  We also push the inodes and
- *                     superblock if we can lock them without sleeping
- *                      and they are not pinned.
- *      SYNC_ATTR    - We need to flush the inodes. Now handled by direct calls
- *                     to xfs_sync_inodes().
- *      SYNC_WAIT    - All the flushes that take place in this call should
- *                     be synchronous.
- *      SYNC_DELWRI  - This tells us to push dirty pages associated with
- *                     inodes.  SYNC_WAIT and SYNC_BDFLUSH are used to
- *                     determine if they should be flushed sync, async, or
- *                     delwri.
- *      SYNC_FSDATA  - This indicates that the caller would like to make
- *                     sure the superblock is safe on disk.  We can ensure
- *                     this by simply making sure the log gets flushed
- *                     if SYNC_BDFLUSH is set, and by actually writing it
- *                     out otherwise.
- *      SYNC_IOWAIT  - The caller wants us to wait for all data I/O to complete
- *                     before we return (including direct I/O). Forms the drain
- *                     side of the write barrier needed to safely quiesce the
- *                     filesystem.
- *
- */
-int
-xfs_sync(
-        xfs_mount_t     *mp,
-        int             flags)
-{
-        int             error;
-        int             last_error = 0;
-        uint            log_flags = XFS_LOG_FORCE;
-
-        ASSERT(!(flags & SYNC_ATTR));
-
-        /*
-         * Get the Quota Manager to flush the dquots.
-         *
-         * If XFS quota support is not enabled or this filesystem
-         * instance does not use quotas XFS_QM_DQSYNC will always
-         * return zero.
-         */
-        error = XFS_QM_DQSYNC(mp, flags);
-        if (error) {
-                /*
-                 * If we got an IO error, we will be shutting down.
-                 * So, there's nothing more for us to do here.
-                 */
-                ASSERT(error != EIO || XFS_FORCED_SHUTDOWN(mp));
-                if (XFS_FORCED_SHUTDOWN(mp))
-                        return XFS_ERROR(error);
-        }
-
-        if (flags & SYNC_IOWAIT)
-                xfs_filestream_flush(mp);
-
-        /*
-         * Sync out the log.  This ensures that the log is periodically
-         * flushed even if there is not enough activity to fill it up.
-         */
-        if (flags & SYNC_WAIT)
-                log_flags |= XFS_LOG_SYNC;
-
-        xfs_log_force(mp, (xfs_lsn_t)0, log_flags);
-
-        if (flags & SYNC_DELWRI) {
-                if (flags & SYNC_BDFLUSH)
-                        xfs_finish_reclaim_all(mp, 1, XFS_IFLUSH_DELWRI_ELSE_ASYNC);
-                else
-                        error = xfs_sync_inodes(mp, flags);
-                /*
-                 * Flushing out dirty data above probably generated more
-                 * log activity, so if this isn't vfs_sync() then flush
-                 * the log again.
-                 */
-                xfs_log_force(mp, 0, log_flags);
-        }
-
-        if (flags & SYNC_FSDATA) {
-                error = xfs_sync_fsdata(mp, flags);
-                if (error)
-                        last_error = error;
-        }
-
-        /*
-         * Now check to see if the log needs a "dummy" transaction.
-         */
-        if (!(flags & SYNC_REMOUNT) && xfs_log_need_covered(mp)) {
-                error = xfs_commit_dummy_trans(mp, log_flags);
-                if (error)
-                        return error;
-        }
-
-        return XFS_ERROR(last_error);
-}
-
-/*
  * Enqueue a work item to be picked up by the vfs xfssyncd thread.
  * Doing this has two advantages:
  * - It saves on stack space, which is tight in certain situations
diff --git a/fs/xfs/linux-2.6/xfs_sync.h b/fs/xfs/linux-2.6/xfs_sync.h
index 2509db021f79..4591dc0c7880 100644
--- a/fs/xfs/linux-2.6/xfs_sync.h
+++ b/fs/xfs/linux-2.6/xfs_sync.h
@@ -28,31 +28,14 @@ typedef struct bhv_vfs_sync_work {
 } bhv_vfs_sync_work_t;
 
 #define SYNC_ATTR               0x0001  /* sync attributes */
-#define SYNC_DELWRI             0x0004  /* look at delayed writes */
-#define SYNC_WAIT               0x0008  /* wait for i/o to complete */
-#define SYNC_BDFLUSH            0x0010  /* BDFLUSH is calling -- don't block */
-#define SYNC_FSDATA             0x0020  /* flush fs data (e.g. superblocks) */
-#define SYNC_REFCACHE           0x0040  /* prune some of the nfs ref cache */
-#define SYNC_REMOUNT            0x0080  /* remount readonly, no dummy LRs */
-#define SYNC_IOWAIT             0x0100  /* wait for all I/O to complete */
-
-/*
- * When remounting a filesystem read-only or freezing the filesystem,
- * we have two phases to execute. This first phase is syncing the data
- * before we quiesce the fielsystem, and the second is flushing all the
- * inodes out after we've waited for all the transactions created by
- * the first phase to complete. The second phase uses SYNC_INODE_QUIESCE
- * to ensure that the inodes are written to their location on disk
- * rather than just existing in transactions in the log. This means
- * after a quiesce there is no log replay required to write the inodes
- * to disk (this is the main difference between a sync and a quiesce).
- */
-#define SYNC_DATA_QUIESCE       (SYNC_DELWRI|SYNC_FSDATA|SYNC_WAIT|SYNC_IOWAIT)
+#define SYNC_DELWRI             0x0002  /* look at delayed writes */
+#define SYNC_WAIT               0x0004  /* wait for i/o to complete */
+#define SYNC_BDFLUSH            0x0008  /* BDFLUSH is calling -- don't block */
+#define SYNC_IOWAIT             0x0010  /* wait for all I/O to complete */
 
 int xfs_syncd_init(struct xfs_mount *mp);
 void xfs_syncd_stop(struct xfs_mount *mp);
 
-int xfs_sync(struct xfs_mount *mp, int flags);
 int xfs_sync_inodes(struct xfs_mount *mp, int flags);
 int xfs_sync_fsdata(struct xfs_mount *mp, int flags);
 
diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index 270f775974e2..db1986a205a9 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -987,14 +987,10 @@ xfs_qm_dqdetach(
 }
 
 /*
- * This is called by VFS_SYNC and flags arg determines the caller,
- * and its motives, as done in xfs_sync.
- *
- * vfs_sync: SYNC_FSDATA|SYNC_ATTR|SYNC_BDFLUSH 0x31
- * syscall sync: SYNC_FSDATA|SYNC_ATTR|SYNC_DELWRI 0x25
- * umountroot : SYNC_WAIT | SYNC_CLOSE | SYNC_ATTR | SYNC_FSDATA
+ * This is called to sync quotas. We can be told to use non-blocking
+ * semantics by either the SYNC_BDFLUSH flag or the absence of the
+ * SYNC_WAIT flag.
  */
-
 int
 xfs_qm_sync(
         xfs_mount_t     *mp,
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index 1256746b249f..58865fe47806 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -431,14 +431,13 @@ xfs_ireclaim(xfs_inode_t *ip)
         xfs_iextract(ip);
 
         /*
-         * Here we do a spurious inode lock in order to coordinate with
-         * xfs_sync().  This is because xfs_sync() references the inodes
-         * in the mount list without taking references on the corresponding
-         * vnodes.  We make that OK here by ensuring that we wait until
-         * the inode is unlocked in xfs_sync() before we go ahead and
-         * free it.  We get both the regular lock and the io lock because
-         * the xfs_sync() code may need to drop the regular one but will
-         * still hold the io lock.
+         * Here we do a spurious inode lock in order to coordinate with inode
+         * cache radix tree lookups.  This is because the lookup can reference
+         * the inodes in the cache without taking references.  We make that OK
+         * here by ensuring that we wait until the inode is unlocked after the
+         * lookup before we go ahead and free it.  We get both the ilock and
+         * the iolock because the code may need to drop the ilock one but will
+         * still hold the iolock.
          */
         xfs_ilock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);