1 files changed, 306 insertions, 141 deletions

diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index 961df0a22c78..a427c638d909 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -44,6 +44,7 @@
 #include "xfs_inode_item.h"
 #include "xfs_rw.h"
 #include "xfs_quota.h"
+#include "xfs_trace.h"
 
 #include <linux/kthread.h>
 #include <linux/freezer.h>
@@ -64,7 +65,6 @@ xfs_inode_ag_lookup(
          * as the tree is sparse and a gang lookup walks to find
          * the number of objects requested.
          */
-        read_lock(&pag->pag_ici_lock);
         if (tag == XFS_ICI_NO_TAG) {
                 nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
                                 (void **)&ip, *first_index, 1);
@@ -73,7 +73,7 @@ xfs_inode_ag_lookup(
                                 (void **)&ip, *first_index, 1, tag);
         }
         if (!nr_found)
-                goto unlock;
+                return NULL;
 
         /*
          * Update the index for the next lookup. Catch overflows
@@ -83,25 +83,21 @@ xfs_inode_ag_lookup(
          */
         *first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
         if (*first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
-                goto unlock;
-
+                return NULL;
         return ip;
-
-unlock:
-        read_unlock(&pag->pag_ici_lock);
-        return NULL;
 }
 
 STATIC int
 xfs_inode_ag_walk(
         struct xfs_mount        *mp,
-        xfs_agnumber_t          ag,
+        struct xfs_perag        *pag,
         int                     (*execute)(struct xfs_inode *ip,
                                            struct xfs_perag *pag, int flags),
         int                     flags,
-        int                     tag)
+        int                     tag,
+        int                     exclusive,
+        int                     *nr_to_scan)
 {
-        struct xfs_perag        *pag = &mp->m_perag[ag];
         uint32_t                first_index;
         int                     last_error = 0;
         int                     skipped;
@@ -113,10 +109,20 @@ restart:
                 int             error = 0;
                 xfs_inode_t     *ip;
 
+                if (exclusive)
+                        write_lock(&pag->pag_ici_lock);
+                else
+                        read_lock(&pag->pag_ici_lock);
                 ip = xfs_inode_ag_lookup(mp, pag, &first_index, tag);
-                if (!ip)
+                if (!ip) {
+                        if (exclusive)
+                                write_unlock(&pag->pag_ici_lock);
+                        else
+                                read_unlock(&pag->pag_ici_lock);
                         break;
+                }
 
+                /* execute releases pag->pag_ici_lock */
                 error = execute(ip, pag, flags);
                 if (error == EAGAIN) {
                         skipped++;
@@ -124,20 +130,17 @@ restart:
                 }
                 if (error)
                         last_error = error;
-                /*
-                 * bail out if the filesystem is corrupted.
-                 */
+
+                /* bail out if the filesystem is corrupted.  */
                 if (error == EFSCORRUPTED)
                         break;
 
-        } while (1);
+        } while ((*nr_to_scan)--);
 
         if (skipped) {
                 delay(1);
                 goto restart;
         }
-
-        xfs_put_perag(mp, pag);
         return last_error;
 }
 
@@ -147,22 +150,37 @@ xfs_inode_ag_iterator(
         int                     (*execute)(struct xfs_inode *ip,
                                            struct xfs_perag *pag, int flags),
         int                     flags,
-        int                     tag)
+        int                     tag,
+        int                     exclusive,
+        int                     *nr_to_scan)
 {
         int                     error = 0;
         int                     last_error = 0;
         xfs_agnumber_t          ag;
+        int                     nr;
 
+        nr = nr_to_scan ? *nr_to_scan : INT_MAX;
         for (ag = 0; ag < mp->m_sb.sb_agcount; ag++) {
-                if (!mp->m_perag[ag].pag_ici_init)
+                struct xfs_perag        *pag;
+
+                pag = xfs_perag_get(mp, ag);
+                if (!pag->pag_ici_init) {
+                        xfs_perag_put(pag);
                         continue;
-                error = xfs_inode_ag_walk(mp, ag, execute, flags, tag);
+                }
+                error = xfs_inode_ag_walk(mp, pag, execute, flags, tag,
+                                                exclusive, &nr);
+                xfs_perag_put(pag);
                 if (error) {
                         last_error = error;
                         if (error == EFSCORRUPTED)
                                 break;
                 }
+                if (nr <= 0)
+                        break;
         }
+        if (nr_to_scan)
+                *nr_to_scan = nr;
         return XFS_ERROR(last_error);
 }
 
@@ -173,30 +191,31 @@ xfs_sync_inode_valid(
         struct xfs_perag        *pag)
 {
         struct inode            *inode = VFS_I(ip);
+        int                     error = EFSCORRUPTED;
 
         /* nothing to sync during shutdown */
-        if (XFS_FORCED_SHUTDOWN(ip->i_mount)) {
-                read_unlock(&pag->pag_ici_lock);
-                return EFSCORRUPTED;
-        }
+        if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+                goto out_unlock;
 
-        /*
-         * If we can't get a reference on the inode, it must be in reclaim.
-         * Leave it for the reclaim code to flush. Also avoid inodes that
-         * haven't been fully initialised.
-         */
-        if (!igrab(inode)) {
-                read_unlock(&pag->pag_ici_lock);
-                return ENOENT;
-        }
-        read_unlock(&pag->pag_ici_lock);
+        /* avoid new or reclaimable inodes. Leave for reclaim code to flush */
+        error = ENOENT;
+        if (xfs_iflags_test(ip, XFS_INEW | XFS_IRECLAIMABLE | XFS_IRECLAIM))
+                goto out_unlock;
+
+        /* If we can't grab the inode, it must on it's way to reclaim. */
+        if (!igrab(inode))
+                goto out_unlock;
 
-        if (is_bad_inode(inode) || xfs_iflags_test(ip, XFS_INEW)) {
+        if (is_bad_inode(inode)) {
                 IRELE(ip);
-                return ENOENT;
+                goto out_unlock;
         }
 
-        return 0;
+        /* inode is valid */
+        error = 0;
+out_unlock:
+        read_unlock(&pag->pag_ici_lock);
+        return error;
 }
 
 STATIC int
@@ -223,7 +242,7 @@ xfs_sync_inode_data(
         }
 
         error = xfs_flush_pages(ip, 0, -1, (flags & SYNC_WAIT) ?
-                                0 : XFS_B_ASYNC, FI_NONE);
+                                0 : XBF_ASYNC, FI_NONE);
         xfs_iunlock(ip, XFS_IOLOCK_SHARED);
 
  out_wait:
@@ -259,8 +278,7 @@ xfs_sync_inode_attr(
                 goto out_unlock;
         }
 
-        error = xfs_iflush(ip, (flags & SYNC_WAIT) ?
-                           XFS_IFLUSH_SYNC : XFS_IFLUSH_DELWRI);
+        error = xfs_iflush(ip, flags);
 
  out_unlock:
         xfs_iunlock(ip, XFS_ILOCK_SHARED);
@@ -281,14 +299,11 @@ xfs_sync_data(
         ASSERT((flags & ~(SYNC_TRYLOCK|SYNC_WAIT)) == 0);
 
         error = xfs_inode_ag_iterator(mp, xfs_sync_inode_data, flags,
-                                      XFS_ICI_NO_TAG);
+                                      XFS_ICI_NO_TAG, 0, NULL);
         if (error)
                 return XFS_ERROR(error);
 
-        xfs_log_force(mp, 0,
-                      (flags & SYNC_WAIT) ?
-                       XFS_LOG_FORCE | XFS_LOG_SYNC :
-                       XFS_LOG_FORCE);
+        xfs_log_force(mp, (flags & SYNC_WAIT) ? XFS_LOG_SYNC : 0);
         return 0;
 }
 
@@ -303,7 +318,7 @@ xfs_sync_attr(
         ASSERT((flags & ~SYNC_WAIT) == 0);
 
         return xfs_inode_ag_iterator(mp, xfs_sync_inode_attr, flags,
-                                     XFS_ICI_NO_TAG);
+                                     XFS_ICI_NO_TAG, 0, NULL);
 }
 
 STATIC int
@@ -314,10 +329,6 @@ xfs_commit_dummy_trans(
         struct xfs_inode        *ip = mp->m_rootip;
         struct xfs_trans        *tp;
         int                     error;
-        int                     log_flags = XFS_LOG_FORCE;
-
-        if (flags & SYNC_WAIT)
-                log_flags |= XFS_LOG_SYNC;
 
         /*
          * Put a dummy transaction in the log to tell recovery
@@ -339,11 +350,11 @@ xfs_commit_dummy_trans(
         xfs_iunlock(ip, XFS_ILOCK_EXCL);
 
         /* the log force ensures this transaction is pushed to disk */
-        xfs_log_force(mp, 0, log_flags);
+        xfs_log_force(mp, (flags & SYNC_WAIT) ? XFS_LOG_SYNC : 0);
         return error;
 }
 
-int
+STATIC int
 xfs_sync_fsdata(
         struct xfs_mount        *mp,
         int                     flags)
@@ -359,7 +370,7 @@ xfs_sync_fsdata(
         if (flags & SYNC_TRYLOCK) {
                 ASSERT(!(flags & SYNC_WAIT));
 
-                bp = xfs_getsb(mp, XFS_BUF_TRYLOCK);
+                bp = xfs_getsb(mp, XBF_TRYLOCK);
                 if (!bp)
                         goto out;
 
@@ -379,7 +390,7 @@ xfs_sync_fsdata(
                  * become pinned in between there and here.
                  */
                 if (XFS_BUF_ISPINNED(bp))
-                        xfs_log_force(mp, 0, XFS_LOG_FORCE);
+                        xfs_log_force(mp, 0);
         }
 
 
@@ -440,9 +451,6 @@ xfs_quiesce_data(
         xfs_sync_data(mp, SYNC_WAIT);
         xfs_qm_sync(mp, SYNC_WAIT);
 
-        /* drop inode references pinned by filestreams */
-        xfs_filestream_flush(mp);
-
         /* write superblock and hoover up shutdown errors */
         error = xfs_sync_fsdata(mp, SYNC_WAIT);
 
@@ -459,16 +467,18 @@ xfs_quiesce_fs(
 {
         int     count = 0, pincount;
 
+        xfs_reclaim_inodes(mp, 0);
         xfs_flush_buftarg(mp->m_ddev_targp, 0);
-        xfs_reclaim_inodes(mp, XFS_IFLUSH_DELWRI_ELSE_ASYNC);
 
         /*
          * This loop must run at least twice.  The first instance of the loop
          * will flush most meta data but that will generate more meta data
          * (typically directory updates).  Which then must be flushed and
-         * logged before we can write the unmount record.
+         * logged before we can write the unmount record. We also so sync
+         * reclaim of inodes to catch any that the above delwri flush skipped.
          */
         do {
+                xfs_reclaim_inodes(mp, SYNC_WAIT);
                 xfs_sync_attr(mp, SYNC_WAIT);
                 pincount = xfs_flush_buftarg(mp->m_ddev_targp, 1);
                 if (!pincount) {
@@ -567,7 +577,7 @@ xfs_flush_inodes(
         igrab(inode);
         xfs_syncd_queue_work(ip->i_mount, inode, xfs_flush_inodes_work, &completion);
         wait_for_completion(&completion);
-        xfs_log_force(ip->i_mount, (xfs_lsn_t)0, XFS_LOG_FORCE|XFS_LOG_SYNC);
+        xfs_log_force(ip->i_mount, XFS_LOG_SYNC);
 }
 
 /*
@@ -583,8 +593,8 @@ xfs_sync_worker(
         int             error;
 
         if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {
-                xfs_log_force(mp, (xfs_lsn_t)0, XFS_LOG_FORCE);
-                xfs_reclaim_inodes(mp, XFS_IFLUSH_DELWRI_ELSE_ASYNC);
+                xfs_log_force(mp, 0);
+                xfs_reclaim_inodes(mp, 0);
                 /* dgc: errors ignored here */
                 error = xfs_qm_sync(mp, SYNC_TRYLOCK);
                 error = xfs_sync_fsdata(mp, SYNC_TRYLOCK);
@@ -605,7 +615,8 @@ xfssyncd(
         set_freezable();
         timeleft = xfs_syncd_centisecs * msecs_to_jiffies(10);
         for (;;) {
-                timeleft = schedule_timeout_interruptible(timeleft);
+                if (list_empty(&mp->m_sync_list))
+                        timeleft = schedule_timeout_interruptible(timeleft);
                 /* swsusp */
                 try_to_freeze();
                 if (kthread_should_stop() && list_empty(&mp->m_sync_list))
@@ -625,8 +636,7 @@ xfssyncd(
                         list_add_tail(&mp->m_sync_work.w_list,
                                         &mp->m_sync_list);
                 }
-                list_for_each_entry_safe(work, n, &mp->m_sync_list, w_list)
-                        list_move(&work->w_list, &tmp);
+                list_splice_init(&mp->m_sync_list, &tmp);
                 spin_unlock(&mp->m_sync_lock);
 
                 list_for_each_entry_safe(work, n, &tmp, w_list) {
@@ -663,67 +673,6 @@ xfs_syncd_stop(
         kthread_stop(mp->m_sync_task);
 }
 
-int
-xfs_reclaim_inode(
-        xfs_inode_t     *ip,
-        int             locked,
-        int             sync_mode)
-{
-        xfs_perag_t     *pag = xfs_get_perag(ip->i_mount, ip->i_ino);
-
-        /* The hash lock here protects a thread in xfs_iget_core from
-         * racing with us on linking the inode back with a vnode.
-         * Once we have the XFS_IRECLAIM flag set it will not touch
-         * us.
-         */
-        write_lock(&pag->pag_ici_lock);
-        spin_lock(&ip->i_flags_lock);
-        if (__xfs_iflags_test(ip, XFS_IRECLAIM) ||
-            !__xfs_iflags_test(ip, XFS_IRECLAIMABLE)) {
-                spin_unlock(&ip->i_flags_lock);
-                write_unlock(&pag->pag_ici_lock);
-                if (locked) {
-                        xfs_ifunlock(ip);
-                        xfs_iunlock(ip, XFS_ILOCK_EXCL);
-                }
-                return -EAGAIN;
-        }
-        __xfs_iflags_set(ip, XFS_IRECLAIM);
-        spin_unlock(&ip->i_flags_lock);
-        write_unlock(&pag->pag_ici_lock);
-        xfs_put_perag(ip->i_mount, pag);
-
-        /*
-         * If the inode is still dirty, then flush it out.  If the inode
-         * is not in the AIL, then it will be OK to flush it delwri as
-         * long as xfs_iflush() does not keep any references to the inode.
-         * We leave that decision up to xfs_iflush() since it has the
-         * knowledge of whether it's OK to simply do a delwri flush of
-         * the inode or whether we need to wait until the inode is
-         * pulled from the AIL.
-         * We get the flush lock regardless, though, just to make sure
-         * we don't free it while it is being flushed.
-         */
-        if (!locked) {
-                xfs_ilock(ip, XFS_ILOCK_EXCL);
-                xfs_iflock(ip);
-        }
-
-        /*
-         * In the case of a forced shutdown we rely on xfs_iflush() to
-         * wait for the inode to be unpinned before returning an error.
-         */
-        if (!is_bad_inode(VFS_I(ip)) && xfs_iflush(ip, sync_mode) == 0) {
-                /* synchronize with xfs_iflush_done */
-                xfs_iflock(ip);
-                xfs_ifunlock(ip);
-        }
-
-        xfs_iunlock(ip, XFS_ILOCK_EXCL);
-        xfs_ireclaim(ip);
-        return 0;
-}
-
 void
 __xfs_inode_set_reclaim_tag(
         struct xfs_perag        *pag,
@@ -732,6 +681,7 @@ __xfs_inode_set_reclaim_tag(
         radix_tree_tag_set(&pag->pag_ici_root,
                            XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino),
                            XFS_ICI_RECLAIM_TAG);
+        pag->pag_ici_reclaimable++;
 }
 
 /*
@@ -743,16 +693,17 @@ void
 xfs_inode_set_reclaim_tag(
         xfs_inode_t     *ip)
 {
-        xfs_mount_t     *mp = ip->i_mount;
-        xfs_perag_t     *pag = xfs_get_perag(mp, ip->i_ino);
+        struct xfs_mount *mp = ip->i_mount;
+        struct xfs_perag *pag;
 
-        read_lock(&pag->pag_ici_lock);
+        pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
+        write_lock(&pag->pag_ici_lock);
         spin_lock(&ip->i_flags_lock);
         __xfs_inode_set_reclaim_tag(pag, ip);
         __xfs_iflags_set(ip, XFS_IRECLAIMABLE);
         spin_unlock(&ip->i_flags_lock);
-        read_unlock(&pag->pag_ici_lock);
-        xfs_put_perag(mp, pag);
+        write_unlock(&pag->pag_ici_lock);
+        xfs_perag_put(pag);
 }
 
 void
@@ -763,22 +714,148 @@ __xfs_inode_clear_reclaim_tag(
 {
         radix_tree_tag_clear(&pag->pag_ici_root,
                         XFS_INO_TO_AGINO(mp, ip->i_ino), XFS_ICI_RECLAIM_TAG);
+        pag->pag_ici_reclaimable--;
 }
 
+/*
+ * Inodes in different states need to be treated differently, and the return
+ * value of xfs_iflush is not sufficient to get this right. The following table
+ * lists the inode states and the reclaim actions necessary for non-blocking
+ * reclaim:
+ *
+ *
+ *      inode state          iflush ret         required action
+ *      ---------------      ----------         ---------------
+ *      bad                     -               reclaim
+ *      shutdown                EIO             unpin and reclaim
+ *      clean, unpinned         0               reclaim
+ *      stale, unpinned         0               reclaim
+ *      clean, pinned(*)        0               requeue
+ *      stale, pinned           EAGAIN          requeue
+ *      dirty, delwri ok        0               requeue
+ *      dirty, delwri blocked   EAGAIN          requeue
+ *      dirty, sync flush       0               reclaim
+ *
+ * (*) dgc: I don't think the clean, pinned state is possible but it gets
+ * handled anyway given the order of checks implemented.
+ *
+ * As can be seen from the table, the return value of xfs_iflush() is not
+ * sufficient to correctly decide the reclaim action here. The checks in
+ * xfs_iflush() might look like duplicates, but they are not.
+ *
+ * Also, because we get the flush lock first, we know that any inode that has
+ * been flushed delwri has had the flush completed by the time we check that
+ * the inode is clean. The clean inode check needs to be done before flushing
+ * the inode delwri otherwise we would loop forever requeuing clean inodes as
+ * we cannot tell apart a successful delwri flush and a clean inode from the
+ * return value of xfs_iflush().
+ *
+ * Note that because the inode is flushed delayed write by background
+ * writeback, the flush lock may already be held here and waiting on it can
+ * result in very long latencies. Hence for sync reclaims, where we wait on the
+ * flush lock, the caller should push out delayed write inodes first before
+ * trying to reclaim them to minimise the amount of time spent waiting. For
+ * background relaim, we just requeue the inode for the next pass.
+ *
+ * Hence the order of actions after gaining the locks should be:
+ *      bad             => reclaim
+ *      shutdown        => unpin and reclaim
+ *      pinned, delwri  => requeue
+ *      pinned, sync    => unpin
+ *      stale           => reclaim
+ *      clean           => reclaim
+ *      dirty, delwri   => flush and requeue
+ *      dirty, sync     => flush, wait and reclaim
+ */
 STATIC int
-xfs_reclaim_inode_now(
+xfs_reclaim_inode(
         struct xfs_inode        *ip,
         struct xfs_perag        *pag,
-        int                     flags)
+        int                     sync_mode)
 {
-        /* ignore if already under reclaim */
-        if (xfs_iflags_test(ip, XFS_IRECLAIM)) {
-                read_unlock(&pag->pag_ici_lock);
+        int     error = 0;
+
+        /*
+         * The radix tree lock here protects a thread in xfs_iget from racing
+         * with us starting reclaim on the inode.  Once we have the
+         * XFS_IRECLAIM flag set it will not touch us.
+         */
+        spin_lock(&ip->i_flags_lock);
+        ASSERT_ALWAYS(__xfs_iflags_test(ip, XFS_IRECLAIMABLE));
+        if (__xfs_iflags_test(ip, XFS_IRECLAIM)) {
+                /* ignore as it is already under reclaim */
+                spin_unlock(&ip->i_flags_lock);
+                write_unlock(&pag->pag_ici_lock);
                 return 0;
         }
-        read_unlock(&pag->pag_ici_lock);
+        __xfs_iflags_set(ip, XFS_IRECLAIM);
+        spin_unlock(&ip->i_flags_lock);
+        write_unlock(&pag->pag_ici_lock);
+
+        xfs_ilock(ip, XFS_ILOCK_EXCL);
+        if (!xfs_iflock_nowait(ip)) {
+                if (!(sync_mode & SYNC_WAIT))
+                        goto out;
+                xfs_iflock(ip);
+        }
+
+        if (is_bad_inode(VFS_I(ip)))
+                goto reclaim;
+        if (XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+                xfs_iunpin_wait(ip);
+                goto reclaim;
+        }
+        if (xfs_ipincount(ip)) {
+                if (!(sync_mode & SYNC_WAIT)) {
+                        xfs_ifunlock(ip);
+                        goto out;
+                }
+                xfs_iunpin_wait(ip);
+        }
+        if (xfs_iflags_test(ip, XFS_ISTALE))
+                goto reclaim;
+        if (xfs_inode_clean(ip))
+                goto reclaim;
+
+        /* Now we have an inode that needs flushing */
+        error = xfs_iflush(ip, sync_mode);
+        if (sync_mode & SYNC_WAIT) {
+                xfs_iflock(ip);
+                goto reclaim;
+        }
+
+        /*
+         * When we have to flush an inode but don't have SYNC_WAIT set, we
+         * flush the inode out using a delwri buffer and wait for the next
+         * call into reclaim to find it in a clean state instead of waiting for
+         * it now. We also don't return errors here - if the error is transient
+         * then the next reclaim pass will flush the inode, and if the error
+         * is permanent then the next sync reclaim will reclaim the inode and
+         * pass on the error.
+         */
+        if (error && error != EAGAIN && !XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+                xfs_fs_cmn_err(CE_WARN, ip->i_mount,
+                        "inode 0x%llx background reclaim flush failed with %d",
+                        (long long)ip->i_ino, error);
+        }
+out:
+        xfs_iflags_clear(ip, XFS_IRECLAIM);
+        xfs_iunlock(ip, XFS_ILOCK_EXCL);
+        /*
+         * We could return EAGAIN here to make reclaim rescan the inode tree in
+         * a short while. However, this just burns CPU time scanning the tree
+         * waiting for IO to complete and xfssyncd never goes back to the idle
+         * state. Instead, return 0 to let the next scheduled background reclaim
+         * attempt to reclaim the inode again.
+         */
+        return 0;
+
+reclaim:
+        xfs_ifunlock(ip);
+        xfs_iunlock(ip, XFS_ILOCK_EXCL);
+        xfs_ireclaim(ip);
+        return error;
 
-        return xfs_reclaim_inode(ip, 0, flags);
 }
 
 int
@@ -786,6 +863,94 @@ xfs_reclaim_inodes(
         xfs_mount_t     *mp,
         int             mode)
 {
-        return xfs_inode_ag_iterator(mp, xfs_reclaim_inode_now, mode,
-                                        XFS_ICI_RECLAIM_TAG);
+        return xfs_inode_ag_iterator(mp, xfs_reclaim_inode, mode,
+                                        XFS_ICI_RECLAIM_TAG, 1, NULL);
+}
+
+/*
+ * Shrinker infrastructure.
+ *
+ * This is all far more complex than it needs to be. It adds a global list of
+ * mounts because the shrinkers can only call a global context. We need to make
+ * the shrinkers pass a context to avoid the need for global state.
+ */
+static LIST_HEAD(xfs_mount_list);
+static struct rw_semaphore xfs_mount_list_lock;
+
+static int
+xfs_reclaim_inode_shrink(
+        int             nr_to_scan,
+        gfp_t           gfp_mask)
+{
+        struct xfs_mount *mp;
+        struct xfs_perag *pag;
+        xfs_agnumber_t  ag;
+        int             reclaimable = 0;
+
+        if (nr_to_scan) {
+                if (!(gfp_mask & __GFP_FS))
+                        return -1;
+
+                down_read(&xfs_mount_list_lock);
+                list_for_each_entry(mp, &xfs_mount_list, m_mplist) {
+                        xfs_inode_ag_iterator(mp, xfs_reclaim_inode, 0,
+                                        XFS_ICI_RECLAIM_TAG, 1, &nr_to_scan);
+                        if (nr_to_scan <= 0)
+                                break;
+                }
+                up_read(&xfs_mount_list_lock);
+        }
+
+        down_read(&xfs_mount_list_lock);
+        list_for_each_entry(mp, &xfs_mount_list, m_mplist) {
+                for (ag = 0; ag < mp->m_sb.sb_agcount; ag++) {
+
+                        pag = xfs_perag_get(mp, ag);
+                        if (!pag->pag_ici_init) {
+                                xfs_perag_put(pag);
+                                continue;
+                        }
+                        reclaimable += pag->pag_ici_reclaimable;
+                        xfs_perag_put(pag);
+                }
+        }
+        up_read(&xfs_mount_list_lock);
+        return reclaimable;
+}
+
+static struct shrinker xfs_inode_shrinker = {
+        .shrink = xfs_reclaim_inode_shrink,
+        .seeks = DEFAULT_SEEKS,
+};
+
+void __init
+xfs_inode_shrinker_init(void)
+{
+        init_rwsem(&xfs_mount_list_lock);
+        register_shrinker(&xfs_inode_shrinker);
+}
+
+void
+xfs_inode_shrinker_destroy(void)
+{
+        ASSERT(list_empty(&xfs_mount_list));
+        unregister_shrinker(&xfs_inode_shrinker);
+}
+
+void
+xfs_inode_shrinker_register(
+        struct xfs_mount        *mp)
+{
+        down_write(&xfs_mount_list_lock);
+        list_add_tail(&mp->m_mplist, &xfs_mount_list);
+        up_write(&xfs_mount_list_lock);
+}
+
+void
+xfs_inode_shrinker_unregister(
+        struct xfs_mount        *mp)
+{
+        down_write(&xfs_mount_list_lock);
+        list_del(&mp->m_mplist);
+        up_write(&xfs_mount_list_lock);
 }