9 files changed, 174 insertions, 62 deletions

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 1db080135c6d..506c24fb5078 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -1073,12 +1073,15 @@ static void ocfs2_write_failure(struct inode *inode,
         for(i = 0; i < wc->w_num_pages; i++) {
                 tmppage = wc->w_pages[i];
 
-                if (ocfs2_should_order_data(inode))
-                        walk_page_buffers(wc->w_handle, page_buffers(tmppage),
-                                          from, to, NULL,
-                                          ocfs2_journal_dirty_data);
-
-                block_commit_write(tmppage, from, to);
+                if (page_has_buffers(tmppage)) {
+                        if (ocfs2_should_order_data(inode))
+                                walk_page_buffers(wc->w_handle,
+                                                  page_buffers(tmppage),
+                                                  from, to, NULL,
+                                                  ocfs2_journal_dirty_data);
+
+                        block_commit_write(tmppage, from, to);
+                }
         }
 }
 
@@ -1901,12 +1904,14 @@ int ocfs2_write_end_nolock(struct address_space *mapping,
                         to = PAGE_CACHE_SIZE;
                 }
 
-                if (ocfs2_should_order_data(inode))
-                        walk_page_buffers(wc->w_handle, page_buffers(tmppage),
-                                          from, to, NULL,
-                                          ocfs2_journal_dirty_data);
-
-                block_commit_write(tmppage, from, to);
+                if (page_has_buffers(tmppage)) {
+                        if (ocfs2_should_order_data(inode))
+                                walk_page_buffers(wc->w_handle,
+                                                  page_buffers(tmppage),
+                                                  from, to, NULL,
+                                                  ocfs2_journal_dirty_data);
+                        block_commit_write(tmppage, from, to);
+                }
         }
 
 out_write_size:
diff --git a/fs/ocfs2/dlm/dlmfs.c b/fs/ocfs2/dlm/dlmfs.c
index e48aba698b77..533a789c3ef8 100644
--- a/fs/ocfs2/dlm/dlmfs.c
+++ b/fs/ocfs2/dlm/dlmfs.c
@@ -267,8 +267,7 @@ static ssize_t dlmfs_file_write(struct file *filp,
         return writelen;
 }
 
-static void dlmfs_init_once(struct kmem_cache *cachep,
-                            void *foo)
+static void dlmfs_init_once(void *foo)
 {
         struct dlmfs_inode_private *ip =
                 (struct dlmfs_inode_private *) foo;
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index e8514e8b6ce8..ec2ed15c3daa 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -1176,7 +1176,7 @@ bail:
         return err;
 }
 
-int ocfs2_permission(struct inode *inode, int mask, struct nameidata *nd)
+int ocfs2_permission(struct inode *inode, int mask)
 {
         int ret;
 
@@ -1766,8 +1766,8 @@ out_inode_unlock:
 out_rw_unlock:
         ocfs2_rw_unlock(inode, 1);
 
-        mutex_unlock(&inode->i_mutex);
 out:
+        mutex_unlock(&inode->i_mutex);
         return ret;
 }
 
diff --git a/fs/ocfs2/file.h b/fs/ocfs2/file.h
index 048ddcaf5c80..1e27b4d017ea 100644
--- a/fs/ocfs2/file.h
+++ b/fs/ocfs2/file.h
@@ -62,8 +62,7 @@ int ocfs2_lock_allocators(struct inode *inode, struct ocfs2_dinode *di,
 int ocfs2_setattr(struct dentry *dentry, struct iattr *attr);
 int ocfs2_getattr(struct vfsmount *mnt, struct dentry *dentry,
                   struct kstat *stat);
-int ocfs2_permission(struct inode *inode, int mask,
-                     struct nameidata *nd);
+int ocfs2_permission(struct inode *inode, int mask);
 
 int ocfs2_should_update_atime(struct inode *inode,
                               struct vfsmount *vfsmnt);
diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c
index a8c19cb3cfdd..7a37240f7a31 100644
--- a/fs/ocfs2/journal.c
+++ b/fs/ocfs2/journal.c
@@ -57,7 +57,7 @@ static int __ocfs2_recovery_thread(void *arg);
 static int ocfs2_commit_cache(struct ocfs2_super *osb);
 static int ocfs2_wait_on_mount(struct ocfs2_super *osb);
 static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb,
-                                      int dirty);
+                                      int dirty, int replayed);
 static int ocfs2_trylock_journal(struct ocfs2_super *osb,
                                  int slot_num);
 static int ocfs2_recover_orphans(struct ocfs2_super *osb,
@@ -562,8 +562,18 @@ done:
         return status;
 }
 
+static void ocfs2_bump_recovery_generation(struct ocfs2_dinode *di)
+{
+        le32_add_cpu(&(di->id1.journal1.ij_recovery_generation), 1);
+}
+
+static u32 ocfs2_get_recovery_generation(struct ocfs2_dinode *di)
+{
+        return le32_to_cpu(di->id1.journal1.ij_recovery_generation);
+}
+
 static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb,
-                                      int dirty)
+                                      int dirty, int replayed)
 {
         int status;
         unsigned int flags;
@@ -593,6 +603,9 @@ static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb,
                 flags &= ~OCFS2_JOURNAL_DIRTY_FL;
         fe->id1.journal1.ij_flags = cpu_to_le32(flags);
 
+        if (replayed)
+                ocfs2_bump_recovery_generation(fe);
+
         status = ocfs2_write_block(osb, bh, journal->j_inode);
         if (status < 0)
                 mlog_errno(status);
@@ -667,7 +680,7 @@ void ocfs2_journal_shutdown(struct ocfs2_super *osb)
                  * Do not toggle if flush was unsuccessful otherwise
                  * will leave dirty metadata in a "clean" journal
                  */
-                status = ocfs2_journal_toggle_dirty(osb, 0);
+                status = ocfs2_journal_toggle_dirty(osb, 0, 0);
                 if (status < 0)
                         mlog_errno(status);
         }
@@ -710,7 +723,7 @@ static void ocfs2_clear_journal_error(struct super_block *sb,
         }
 }
 
-int ocfs2_journal_load(struct ocfs2_journal *journal, int local)
+int ocfs2_journal_load(struct ocfs2_journal *journal, int local, int replayed)
 {
         int status = 0;
         struct ocfs2_super *osb;
@@ -729,7 +742,7 @@ int ocfs2_journal_load(struct ocfs2_journal *journal, int local)
 
         ocfs2_clear_journal_error(osb->sb, journal->j_journal, osb->slot_num);
 
-        status = ocfs2_journal_toggle_dirty(osb, 1);
+        status = ocfs2_journal_toggle_dirty(osb, 1, replayed);
         if (status < 0) {
                 mlog_errno(status);
                 goto done;
@@ -771,7 +784,7 @@ int ocfs2_journal_wipe(struct ocfs2_journal *journal, int full)
                 goto bail;
         }
 
-        status = ocfs2_journal_toggle_dirty(journal->j_osb, 0);
+        status = ocfs2_journal_toggle_dirty(journal->j_osb, 0, 0);
         if (status < 0)
                 mlog_errno(status);
 
@@ -1034,6 +1047,12 @@ restart:
         spin_unlock(&osb->osb_lock);
         mlog(0, "All nodes recovered\n");
 
+        /* Refresh all journal recovery generations from disk */
+        status = ocfs2_check_journals_nolocks(osb);
+        status = (status == -EROFS) ? 0 : status;
+        if (status < 0)
+                mlog_errno(status);
+
         ocfs2_super_unlock(osb, 1);
 
         /* We always run recovery on our own orphan dir - the dead
@@ -1096,6 +1115,42 @@ out:
         mlog_exit_void();
 }
 
+static int ocfs2_read_journal_inode(struct ocfs2_super *osb,
+                                    int slot_num,
+                                    struct buffer_head **bh,
+                                    struct inode **ret_inode)
+{
+        int status = -EACCES;
+        struct inode *inode = NULL;
+
+        BUG_ON(slot_num >= osb->max_slots);
+
+        inode = ocfs2_get_system_file_inode(osb, JOURNAL_SYSTEM_INODE,
+                                            slot_num);
+        if (!inode || is_bad_inode(inode)) {
+                mlog_errno(status);
+                goto bail;
+        }
+        SET_INODE_JOURNAL(inode);
+
+        status = ocfs2_read_block(osb, OCFS2_I(inode)->ip_blkno, bh, 0, inode);
+        if (status < 0) {
+                mlog_errno(status);
+                goto bail;
+        }
+
+        status = 0;
+
+bail:
+        if (inode) {
+                if (status || !ret_inode)
+                        iput(inode);
+                else
+                        *ret_inode = inode;
+        }
+        return status;
+}
+
 /* Does the actual journal replay and marks the journal inode as
  * clean. Will only replay if the journal inode is marked dirty. */
 static int ocfs2_replay_journal(struct ocfs2_super *osb,
@@ -1109,22 +1164,36 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
         struct ocfs2_dinode *fe;
         journal_t *journal = NULL;
         struct buffer_head *bh = NULL;
+        u32 slot_reco_gen;
 
-        inode = ocfs2_get_system_file_inode(osb, JOURNAL_SYSTEM_INODE,
-                                            slot_num);
-        if (inode == NULL) {
-                status = -EACCES;
+        status = ocfs2_read_journal_inode(osb, slot_num, &bh, &inode);
+        if (status) {
                 mlog_errno(status);
                 goto done;
         }
-        if (is_bad_inode(inode)) {
-                status = -EACCES;
-                iput(inode);
-                inode = NULL;
-                mlog_errno(status);
+
+        fe = (struct ocfs2_dinode *)bh->b_data;
+        slot_reco_gen = ocfs2_get_recovery_generation(fe);
+        brelse(bh);
+        bh = NULL;
+
+        /*
+         * As the fs recovery is asynchronous, there is a small chance that
+         * another node mounted (and recovered) the slot before the recovery
+         * thread could get the lock. To handle that, we dirty read the journal
+         * inode for that slot to get the recovery generation. If it is
+         * different than what we expected, the slot has been recovered.
+         * If not, it needs recovery.
+         */
+        if (osb->slot_recovery_generations[slot_num] != slot_reco_gen) {
+                mlog(0, "Slot %u already recovered (old/new=%u/%u)\n", slot_num,
+                     osb->slot_recovery_generations[slot_num], slot_reco_gen);
+                osb->slot_recovery_generations[slot_num] = slot_reco_gen;
+                status = -EBUSY;
                 goto done;
         }
-        SET_INODE_JOURNAL(inode);
+
+        /* Continue with recovery as the journal has not yet been recovered */
 
         status = ocfs2_inode_lock_full(inode, &bh, 1, OCFS2_META_LOCK_RECOVERY);
         if (status < 0) {
@@ -1138,9 +1207,12 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
         fe = (struct ocfs2_dinode *) bh->b_data;
 
         flags = le32_to_cpu(fe->id1.journal1.ij_flags);
+        slot_reco_gen = ocfs2_get_recovery_generation(fe);
 
         if (!(flags & OCFS2_JOURNAL_DIRTY_FL)) {
                 mlog(0, "No recovery required for node %d\n", node_num);
+                /* Refresh recovery generation for the slot */
+                osb->slot_recovery_generations[slot_num] = slot_reco_gen;
                 goto done;
         }
 
@@ -1188,6 +1260,11 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
         flags &= ~OCFS2_JOURNAL_DIRTY_FL;
         fe->id1.journal1.ij_flags = cpu_to_le32(flags);
 
+        /* Increment recovery generation to indicate successful recovery */
+        ocfs2_bump_recovery_generation(fe);
+        osb->slot_recovery_generations[slot_num] =
+                                        ocfs2_get_recovery_generation(fe);
+
         status = ocfs2_write_block(osb, bh, inode);
         if (status < 0)
                 mlog_errno(status);
@@ -1252,6 +1329,13 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
 
         status = ocfs2_replay_journal(osb, node_num, slot_num);
         if (status < 0) {
+                if (status == -EBUSY) {
+                        mlog(0, "Skipping recovery for slot %u (node %u) "
+                             "as another node has recovered it\n", slot_num,
+                             node_num);
+                        status = 0;
+                        goto done;
+                }
                 mlog_errno(status);
                 goto done;
         }
@@ -1334,12 +1418,29 @@ int ocfs2_mark_dead_nodes(struct ocfs2_super *osb)
 {
         unsigned int node_num;
         int status, i;
+        struct buffer_head *bh = NULL;
+        struct ocfs2_dinode *di;
 
         /* This is called with the super block cluster lock, so we
          * know that the slot map can't change underneath us. */
 
         spin_lock(&osb->osb_lock);
         for (i = 0; i < osb->max_slots; i++) {
+                /* Read journal inode to get the recovery generation */
+                status = ocfs2_read_journal_inode(osb, i, &bh, NULL);
+                if (status) {
+                        mlog_errno(status);
+                        goto bail;
+                }
+                di = (struct ocfs2_dinode *)bh->b_data;
+                osb->slot_recovery_generations[i] =
+                                        ocfs2_get_recovery_generation(di);
+                brelse(bh);
+                bh = NULL;
+
+                mlog(0, "Slot %u recovery generation is %u\n", i,
+                     osb->slot_recovery_generations[i]);
+
                 if (i == osb->slot_num)
                         continue;
 
@@ -1603,49 +1704,41 @@ static int ocfs2_commit_thread(void *arg)
         return 0;
 }
 
-/* Look for a dirty journal without taking any cluster locks. Used for
- * hard readonly access to determine whether the file system journals
- * require recovery. */
+/* Reads all the journal inodes without taking any cluster locks. Used
+ * for hard readonly access to determine whether any journal requires
+ * recovery. Also used to refresh the recovery generation numbers after
+ * a journal has been recovered by another node.
+ */
 int ocfs2_check_journals_nolocks(struct ocfs2_super *osb)
 {
         int ret = 0;
         unsigned int slot;
-        struct buffer_head *di_bh;
+        struct buffer_head *di_bh = NULL;
         struct ocfs2_dinode *di;
-        struct inode *journal = NULL;
+        int journal_dirty = 0;
 
         for(slot = 0; slot < osb->max_slots; slot++) {
-                journal = ocfs2_get_system_file_inode(osb,
-                                                      JOURNAL_SYSTEM_INODE,
-                                                      slot);
-                if (!journal || is_bad_inode(journal)) {
-                        ret = -EACCES;
-                        mlog_errno(ret);
-                        goto out;
-                }
-
-                di_bh = NULL;
-                ret = ocfs2_read_block(osb, OCFS2_I(journal)->ip_blkno, &di_bh,
-                                       0, journal);
-                if (ret < 0) {
+                ret = ocfs2_read_journal_inode(osb, slot, &di_bh, NULL);
+                if (ret) {
                         mlog_errno(ret);
                         goto out;
                 }
 
                 di = (struct ocfs2_dinode *) di_bh->b_data;
 
+                osb->slot_recovery_generations[slot] =
+                                        ocfs2_get_recovery_generation(di);
+
                 if (le32_to_cpu(di->id1.journal1.ij_flags) &
                     OCFS2_JOURNAL_DIRTY_FL)
-                        ret = -EROFS;
+                        journal_dirty = 1;
 
                 brelse(di_bh);
-                if (ret)
-                        break;
+                di_bh = NULL;
         }
 
 out:
-        if (journal)
-                iput(journal);
-
+        if (journal_dirty)
+                ret = -EROFS;
         return ret;
 }
diff --git a/fs/ocfs2/journal.h b/fs/ocfs2/journal.h
index db82be2532ed..2178ebffa05f 100644
--- a/fs/ocfs2/journal.h
+++ b/fs/ocfs2/journal.h
@@ -161,7 +161,8 @@ int    ocfs2_journal_init(struct ocfs2_journal *journal,
 void   ocfs2_journal_shutdown(struct ocfs2_super *osb);
 int    ocfs2_journal_wipe(struct ocfs2_journal *journal,
                           int full);
-int    ocfs2_journal_load(struct ocfs2_journal *journal, int local);
+int    ocfs2_journal_load(struct ocfs2_journal *journal, int local,
+                          int replayed);
 int    ocfs2_check_journals_nolocks(struct ocfs2_super *osb);
 void   ocfs2_recovery_thread(struct ocfs2_super *osb,
                              int node_num);
diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h
index 1cb814be8ef1..7f625f2b1117 100644
--- a/fs/ocfs2/ocfs2.h
+++ b/fs/ocfs2/ocfs2.h
@@ -204,6 +204,8 @@ struct ocfs2_super
 
         struct ocfs2_slot_info *slot_info;
 
+        u32 *slot_recovery_generations;
+
         spinlock_t node_map_lock;
 
         u64 root_blkno;
diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h
index 3f1945177629..4f619850ccf7 100644
--- a/fs/ocfs2/ocfs2_fs.h
+++ b/fs/ocfs2/ocfs2_fs.h
@@ -660,7 +660,10 @@ struct ocfs2_dinode {
                 struct {                /* Info for journal system
                                            inodes */
                         __le32 ij_flags;        /* Mounted, version, etc. */
-                        __le32 ij_pad;
+                        __le32 ij_recovery_generation; /* Incremented when the
+                                                          journal is recovered
+                                                          after an unclean
+                                                          shutdown */
                 } journal1;
         } id1;                          /* Inode type dependant 1 */
 /*C0*/  union {
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index ccecfe5094fa..88255d3f52b4 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -1118,7 +1118,7 @@ bail:
         return status;
 }
 
-static void ocfs2_inode_init_once(struct kmem_cache *cachep, void *data)
+static void ocfs2_inode_init_once(void *data)
 {
         struct ocfs2_inode_info *oi = data;
 
@@ -1442,6 +1442,15 @@ static int ocfs2_initialize_super(struct super_block *sb,
         }
         mlog(0, "max_slots for this device: %u\n", osb->max_slots);
 
+        osb->slot_recovery_generations =
+                kcalloc(osb->max_slots, sizeof(*osb->slot_recovery_generations),
+                        GFP_KERNEL);
+        if (!osb->slot_recovery_generations) {
+                status = -ENOMEM;
+                mlog_errno(status);
+                goto bail;
+        }
+
         init_waitqueue_head(&osb->osb_wipe_event);
         osb->osb_orphan_wipes = kcalloc(osb->max_slots,
                                         sizeof(*osb->osb_orphan_wipes),
@@ -1703,7 +1712,7 @@ static int ocfs2_check_volume(struct ocfs2_super *osb)
         local = ocfs2_mount_local(osb);
 
         /* will play back anything left in the journal. */
-        status = ocfs2_journal_load(osb->journal, local);
+        status = ocfs2_journal_load(osb->journal, local, dirty);
         if (status < 0) {
                 mlog(ML_ERROR, "ocfs2 journal load failed! %d\n", status);
                 goto finally;
@@ -1768,6 +1777,7 @@ static void ocfs2_delete_osb(struct ocfs2_super *osb)
         ocfs2_free_slot_info(osb);
 
         kfree(osb->osb_orphan_wipes);
+        kfree(osb->slot_recovery_generations);
         /* FIXME
          * This belongs in journal shutdown, but because we have to
          * allocate osb->journal at the start of ocfs2_initalize_osb(),