Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / fs / nilfs2 / super.c
diff --git a/kernel/fs/nilfs2/super.c b/kernel/fs/nilfs2/super.c
new file mode 100644 (file)
index 0000000..f47585b
--- /dev/null
@@ -0,0 +1,1486 @@
+/*
+ * super.c - NILFS module and super block management.
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * 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; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will 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 to the Free Software
+ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ *
+ * Written by Ryusuke Konishi <ryusuke@osrg.net>
+ */
+/*
+ *  linux/fs/ext2/super.c
+ *
+ * Copyright (C) 1992, 1993, 1994, 1995
+ * Remy Card (card@masi.ibp.fr)
+ * Laboratoire MASI - Institut Blaise Pascal
+ * Universite Pierre et Marie Curie (Paris VI)
+ *
+ *  from
+ *
+ *  linux/fs/minix/inode.c
+ *
+ *  Copyright (C) 1991, 1992  Linus Torvalds
+ *
+ *  Big-endian to little-endian byte-swapping/bitmaps by
+ *        David S. Miller (davem@caip.rutgers.edu), 1995
+ */
+
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/slab.h>
+#include <linux/init.h>
+#include <linux/blkdev.h>
+#include <linux/parser.h>
+#include <linux/crc32.h>
+#include <linux/vfs.h>
+#include <linux/writeback.h>
+#include <linux/seq_file.h>
+#include <linux/mount.h>
+#include "nilfs.h"
+#include "export.h"
+#include "mdt.h"
+#include "alloc.h"
+#include "btree.h"
+#include "btnode.h"
+#include "page.h"
+#include "cpfile.h"
+#include "sufile.h" /* nilfs_sufile_resize(), nilfs_sufile_set_alloc_range() */
+#include "ifile.h"
+#include "dat.h"
+#include "segment.h"
+#include "segbuf.h"
+
+MODULE_AUTHOR("NTT Corp.");
+MODULE_DESCRIPTION("A New Implementation of the Log-structured Filesystem "
+                  "(NILFS)");
+MODULE_LICENSE("GPL");
+
+static struct kmem_cache *nilfs_inode_cachep;
+struct kmem_cache *nilfs_transaction_cachep;
+struct kmem_cache *nilfs_segbuf_cachep;
+struct kmem_cache *nilfs_btree_path_cache;
+
+static int nilfs_setup_super(struct super_block *sb, int is_mount);
+static int nilfs_remount(struct super_block *sb, int *flags, char *data);
+
+static void nilfs_set_error(struct super_block *sb)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_super_block **sbp;
+
+       down_write(&nilfs->ns_sem);
+       if (!(nilfs->ns_mount_state & NILFS_ERROR_FS)) {
+               nilfs->ns_mount_state |= NILFS_ERROR_FS;
+               sbp = nilfs_prepare_super(sb, 0);
+               if (likely(sbp)) {
+                       sbp[0]->s_state |= cpu_to_le16(NILFS_ERROR_FS);
+                       if (sbp[1])
+                               sbp[1]->s_state |= cpu_to_le16(NILFS_ERROR_FS);
+                       nilfs_commit_super(sb, NILFS_SB_COMMIT_ALL);
+               }
+       }
+       up_write(&nilfs->ns_sem);
+}
+
+/**
+ * nilfs_error() - report failure condition on a filesystem
+ *
+ * nilfs_error() sets an ERROR_FS flag on the superblock as well as
+ * reporting an error message.  It should be called when NILFS detects
+ * incoherences or defects of meta data on disk.  As for sustainable
+ * errors such as a single-shot I/O error, nilfs_warning() or the printk()
+ * function should be used instead.
+ *
+ * The segment constructor must not call this function because it can
+ * kill itself.
+ */
+void nilfs_error(struct super_block *sb, const char *function,
+                const char *fmt, ...)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct va_format vaf;
+       va_list args;
+
+       va_start(args, fmt);
+
+       vaf.fmt = fmt;
+       vaf.va = &args;
+
+       printk(KERN_CRIT "NILFS error (device %s): %s: %pV\n",
+              sb->s_id, function, &vaf);
+
+       va_end(args);
+
+       if (!(sb->s_flags & MS_RDONLY)) {
+               nilfs_set_error(sb);
+
+               if (nilfs_test_opt(nilfs, ERRORS_RO)) {
+                       printk(KERN_CRIT "Remounting filesystem read-only\n");
+                       sb->s_flags |= MS_RDONLY;
+               }
+       }
+
+       if (nilfs_test_opt(nilfs, ERRORS_PANIC))
+               panic("NILFS (device %s): panic forced after error\n",
+                     sb->s_id);
+}
+
+void nilfs_warning(struct super_block *sb, const char *function,
+                  const char *fmt, ...)
+{
+       struct va_format vaf;
+       va_list args;
+
+       va_start(args, fmt);
+
+       vaf.fmt = fmt;
+       vaf.va = &args;
+
+       printk(KERN_WARNING "NILFS warning (device %s): %s: %pV\n",
+              sb->s_id, function, &vaf);
+
+       va_end(args);
+}
+
+
+struct inode *nilfs_alloc_inode(struct super_block *sb)
+{
+       struct nilfs_inode_info *ii;
+
+       ii = kmem_cache_alloc(nilfs_inode_cachep, GFP_NOFS);
+       if (!ii)
+               return NULL;
+       ii->i_bh = NULL;
+       ii->i_state = 0;
+       ii->i_cno = 0;
+       ii->vfs_inode.i_version = 1;
+       nilfs_mapping_init(&ii->i_btnode_cache, &ii->vfs_inode);
+       return &ii->vfs_inode;
+}
+
+static void nilfs_i_callback(struct rcu_head *head)
+{
+       struct inode *inode = container_of(head, struct inode, i_rcu);
+       struct nilfs_mdt_info *mdi = NILFS_MDT(inode);
+
+       if (mdi) {
+               kfree(mdi->mi_bgl); /* kfree(NULL) is safe */
+               kfree(mdi);
+       }
+       kmem_cache_free(nilfs_inode_cachep, NILFS_I(inode));
+}
+
+void nilfs_destroy_inode(struct inode *inode)
+{
+       call_rcu(&inode->i_rcu, nilfs_i_callback);
+}
+
+static int nilfs_sync_super(struct super_block *sb, int flag)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       int err;
+
+ retry:
+       set_buffer_dirty(nilfs->ns_sbh[0]);
+       if (nilfs_test_opt(nilfs, BARRIER)) {
+               err = __sync_dirty_buffer(nilfs->ns_sbh[0],
+                                         WRITE_SYNC | WRITE_FLUSH_FUA);
+       } else {
+               err = sync_dirty_buffer(nilfs->ns_sbh[0]);
+       }
+
+       if (unlikely(err)) {
+               printk(KERN_ERR
+                      "NILFS: unable to write superblock (err=%d)\n", err);
+               if (err == -EIO && nilfs->ns_sbh[1]) {
+                       /*
+                        * sbp[0] points to newer log than sbp[1],
+                        * so copy sbp[0] to sbp[1] to take over sbp[0].
+                        */
+                       memcpy(nilfs->ns_sbp[1], nilfs->ns_sbp[0],
+                              nilfs->ns_sbsize);
+                       nilfs_fall_back_super_block(nilfs);
+                       goto retry;
+               }
+       } else {
+               struct nilfs_super_block *sbp = nilfs->ns_sbp[0];
+
+               nilfs->ns_sbwcount++;
+
+               /*
+                * The latest segment becomes trailable from the position
+                * written in superblock.
+                */
+               clear_nilfs_discontinued(nilfs);
+
+               /* update GC protection for recent segments */
+               if (nilfs->ns_sbh[1]) {
+                       if (flag == NILFS_SB_COMMIT_ALL) {
+                               set_buffer_dirty(nilfs->ns_sbh[1]);
+                               if (sync_dirty_buffer(nilfs->ns_sbh[1]) < 0)
+                                       goto out;
+                       }
+                       if (le64_to_cpu(nilfs->ns_sbp[1]->s_last_cno) <
+                           le64_to_cpu(nilfs->ns_sbp[0]->s_last_cno))
+                               sbp = nilfs->ns_sbp[1];
+               }
+
+               spin_lock(&nilfs->ns_last_segment_lock);
+               nilfs->ns_prot_seq = le64_to_cpu(sbp->s_last_seq);
+               spin_unlock(&nilfs->ns_last_segment_lock);
+       }
+ out:
+       return err;
+}
+
+void nilfs_set_log_cursor(struct nilfs_super_block *sbp,
+                         struct the_nilfs *nilfs)
+{
+       sector_t nfreeblocks;
+
+       /* nilfs->ns_sem must be locked by the caller. */
+       nilfs_count_free_blocks(nilfs, &nfreeblocks);
+       sbp->s_free_blocks_count = cpu_to_le64(nfreeblocks);
+
+       spin_lock(&nilfs->ns_last_segment_lock);
+       sbp->s_last_seq = cpu_to_le64(nilfs->ns_last_seq);
+       sbp->s_last_pseg = cpu_to_le64(nilfs->ns_last_pseg);
+       sbp->s_last_cno = cpu_to_le64(nilfs->ns_last_cno);
+       spin_unlock(&nilfs->ns_last_segment_lock);
+}
+
+struct nilfs_super_block **nilfs_prepare_super(struct super_block *sb,
+                                              int flip)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_super_block **sbp = nilfs->ns_sbp;
+
+       /* nilfs->ns_sem must be locked by the caller. */
+       if (sbp[0]->s_magic != cpu_to_le16(NILFS_SUPER_MAGIC)) {
+               if (sbp[1] &&
+                   sbp[1]->s_magic == cpu_to_le16(NILFS_SUPER_MAGIC)) {
+                       memcpy(sbp[0], sbp[1], nilfs->ns_sbsize);
+               } else {
+                       printk(KERN_CRIT "NILFS: superblock broke on dev %s\n",
+                              sb->s_id);
+                       return NULL;
+               }
+       } else if (sbp[1] &&
+                  sbp[1]->s_magic != cpu_to_le16(NILFS_SUPER_MAGIC)) {
+                       memcpy(sbp[1], sbp[0], nilfs->ns_sbsize);
+       }
+
+       if (flip && sbp[1])
+               nilfs_swap_super_block(nilfs);
+
+       return sbp;
+}
+
+int nilfs_commit_super(struct super_block *sb, int flag)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_super_block **sbp = nilfs->ns_sbp;
+       time_t t;
+
+       /* nilfs->ns_sem must be locked by the caller. */
+       t = get_seconds();
+       nilfs->ns_sbwtime = t;
+       sbp[0]->s_wtime = cpu_to_le64(t);
+       sbp[0]->s_sum = 0;
+       sbp[0]->s_sum = cpu_to_le32(crc32_le(nilfs->ns_crc_seed,
+                                            (unsigned char *)sbp[0],
+                                            nilfs->ns_sbsize));
+       if (flag == NILFS_SB_COMMIT_ALL && sbp[1]) {
+               sbp[1]->s_wtime = sbp[0]->s_wtime;
+               sbp[1]->s_sum = 0;
+               sbp[1]->s_sum = cpu_to_le32(crc32_le(nilfs->ns_crc_seed,
+                                           (unsigned char *)sbp[1],
+                                           nilfs->ns_sbsize));
+       }
+       clear_nilfs_sb_dirty(nilfs);
+       nilfs->ns_flushed_device = 1;
+       /* make sure store to ns_flushed_device cannot be reordered */
+       smp_wmb();
+       return nilfs_sync_super(sb, flag);
+}
+
+/**
+ * nilfs_cleanup_super() - write filesystem state for cleanup
+ * @sb: super block instance to be unmounted or degraded to read-only
+ *
+ * This function restores state flags in the on-disk super block.
+ * This will set "clean" flag (i.e. NILFS_VALID_FS) unless the
+ * filesystem was not clean previously.
+ */
+int nilfs_cleanup_super(struct super_block *sb)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_super_block **sbp;
+       int flag = NILFS_SB_COMMIT;
+       int ret = -EIO;
+
+       sbp = nilfs_prepare_super(sb, 0);
+       if (sbp) {
+               sbp[0]->s_state = cpu_to_le16(nilfs->ns_mount_state);
+               nilfs_set_log_cursor(sbp[0], nilfs);
+               if (sbp[1] && sbp[0]->s_last_cno == sbp[1]->s_last_cno) {
+                       /*
+                        * make the "clean" flag also to the opposite
+                        * super block if both super blocks point to
+                        * the same checkpoint.
+                        */
+                       sbp[1]->s_state = sbp[0]->s_state;
+                       flag = NILFS_SB_COMMIT_ALL;
+               }
+               ret = nilfs_commit_super(sb, flag);
+       }
+       return ret;
+}
+
+/**
+ * nilfs_move_2nd_super - relocate secondary super block
+ * @sb: super block instance
+ * @sb2off: new offset of the secondary super block (in bytes)
+ */
+static int nilfs_move_2nd_super(struct super_block *sb, loff_t sb2off)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct buffer_head *nsbh;
+       struct nilfs_super_block *nsbp;
+       sector_t blocknr, newblocknr;
+       unsigned long offset;
+       int sb2i = -1;  /* array index of the secondary superblock */
+       int ret = 0;
+
+       /* nilfs->ns_sem must be locked by the caller. */
+       if (nilfs->ns_sbh[1] &&
+           nilfs->ns_sbh[1]->b_blocknr > nilfs->ns_first_data_block) {
+               sb2i = 1;
+               blocknr = nilfs->ns_sbh[1]->b_blocknr;
+       } else if (nilfs->ns_sbh[0]->b_blocknr > nilfs->ns_first_data_block) {
+               sb2i = 0;
+               blocknr = nilfs->ns_sbh[0]->b_blocknr;
+       }
+       if (sb2i >= 0 && (u64)blocknr << nilfs->ns_blocksize_bits == sb2off)
+               goto out;  /* super block location is unchanged */
+
+       /* Get new super block buffer */
+       newblocknr = sb2off >> nilfs->ns_blocksize_bits;
+       offset = sb2off & (nilfs->ns_blocksize - 1);
+       nsbh = sb_getblk(sb, newblocknr);
+       if (!nsbh) {
+               printk(KERN_WARNING
+                      "NILFS warning: unable to move secondary superblock "
+                      "to block %llu\n", (unsigned long long)newblocknr);
+               ret = -EIO;
+               goto out;
+       }
+       nsbp = (void *)nsbh->b_data + offset;
+       memset(nsbp, 0, nilfs->ns_blocksize);
+
+       if (sb2i >= 0) {
+               memcpy(nsbp, nilfs->ns_sbp[sb2i], nilfs->ns_sbsize);
+               brelse(nilfs->ns_sbh[sb2i]);
+               nilfs->ns_sbh[sb2i] = nsbh;
+               nilfs->ns_sbp[sb2i] = nsbp;
+       } else if (nilfs->ns_sbh[0]->b_blocknr < nilfs->ns_first_data_block) {
+               /* secondary super block will be restored to index 1 */
+               nilfs->ns_sbh[1] = nsbh;
+               nilfs->ns_sbp[1] = nsbp;
+       } else {
+               brelse(nsbh);
+       }
+out:
+       return ret;
+}
+
+/**
+ * nilfs_resize_fs - resize the filesystem
+ * @sb: super block instance
+ * @newsize: new size of the filesystem (in bytes)
+ */
+int nilfs_resize_fs(struct super_block *sb, __u64 newsize)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_super_block **sbp;
+       __u64 devsize, newnsegs;
+       loff_t sb2off;
+       int ret;
+
+       ret = -ERANGE;
+       devsize = i_size_read(sb->s_bdev->bd_inode);
+       if (newsize > devsize)
+               goto out;
+
+       /*
+        * Write lock is required to protect some functions depending
+        * on the number of segments, the number of reserved segments,
+        * and so forth.
+        */
+       down_write(&nilfs->ns_segctor_sem);
+
+       sb2off = NILFS_SB2_OFFSET_BYTES(newsize);
+       newnsegs = sb2off >> nilfs->ns_blocksize_bits;
+       do_div(newnsegs, nilfs->ns_blocks_per_segment);
+
+       ret = nilfs_sufile_resize(nilfs->ns_sufile, newnsegs);
+       up_write(&nilfs->ns_segctor_sem);
+       if (ret < 0)
+               goto out;
+
+       ret = nilfs_construct_segment(sb);
+       if (ret < 0)
+               goto out;
+
+       down_write(&nilfs->ns_sem);
+       nilfs_move_2nd_super(sb, sb2off);
+       ret = -EIO;
+       sbp = nilfs_prepare_super(sb, 0);
+       if (likely(sbp)) {
+               nilfs_set_log_cursor(sbp[0], nilfs);
+               /*
+                * Drop NILFS_RESIZE_FS flag for compatibility with
+                * mount-time resize which may be implemented in a
+                * future release.
+                */
+               sbp[0]->s_state = cpu_to_le16(le16_to_cpu(sbp[0]->s_state) &
+                                             ~NILFS_RESIZE_FS);
+               sbp[0]->s_dev_size = cpu_to_le64(newsize);
+               sbp[0]->s_nsegments = cpu_to_le64(nilfs->ns_nsegments);
+               if (sbp[1])
+                       memcpy(sbp[1], sbp[0], nilfs->ns_sbsize);
+               ret = nilfs_commit_super(sb, NILFS_SB_COMMIT_ALL);
+       }
+       up_write(&nilfs->ns_sem);
+
+       /*
+        * Reset the range of allocatable segments last.  This order
+        * is important in the case of expansion because the secondary
+        * superblock must be protected from log write until migration
+        * completes.
+        */
+       if (!ret)
+               nilfs_sufile_set_alloc_range(nilfs->ns_sufile, 0, newnsegs - 1);
+out:
+       return ret;
+}
+
+static void nilfs_put_super(struct super_block *sb)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+
+       nilfs_detach_log_writer(sb);
+
+       if (!(sb->s_flags & MS_RDONLY)) {
+               down_write(&nilfs->ns_sem);
+               nilfs_cleanup_super(sb);
+               up_write(&nilfs->ns_sem);
+       }
+
+       iput(nilfs->ns_sufile);
+       iput(nilfs->ns_cpfile);
+       iput(nilfs->ns_dat);
+
+       destroy_nilfs(nilfs);
+       sb->s_fs_info = NULL;
+}
+
+static int nilfs_sync_fs(struct super_block *sb, int wait)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_super_block **sbp;
+       int err = 0;
+
+       /* This function is called when super block should be written back */
+       if (wait)
+               err = nilfs_construct_segment(sb);
+
+       down_write(&nilfs->ns_sem);
+       if (nilfs_sb_dirty(nilfs)) {
+               sbp = nilfs_prepare_super(sb, nilfs_sb_will_flip(nilfs));
+               if (likely(sbp)) {
+                       nilfs_set_log_cursor(sbp[0], nilfs);
+                       nilfs_commit_super(sb, NILFS_SB_COMMIT);
+               }
+       }
+       up_write(&nilfs->ns_sem);
+
+       if (!err)
+               err = nilfs_flush_device(nilfs);
+
+       return err;
+}
+
+int nilfs_attach_checkpoint(struct super_block *sb, __u64 cno, int curr_mnt,
+                           struct nilfs_root **rootp)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_root *root;
+       struct nilfs_checkpoint *raw_cp;
+       struct buffer_head *bh_cp;
+       int err = -ENOMEM;
+
+       root = nilfs_find_or_create_root(
+               nilfs, curr_mnt ? NILFS_CPTREE_CURRENT_CNO : cno);
+       if (!root)
+               return err;
+
+       if (root->ifile)
+               goto reuse; /* already attached checkpoint */
+
+       down_read(&nilfs->ns_segctor_sem);
+       err = nilfs_cpfile_get_checkpoint(nilfs->ns_cpfile, cno, 0, &raw_cp,
+                                         &bh_cp);
+       up_read(&nilfs->ns_segctor_sem);
+       if (unlikely(err)) {
+               if (err == -ENOENT || err == -EINVAL) {
+                       printk(KERN_ERR
+                              "NILFS: Invalid checkpoint "
+                              "(checkpoint number=%llu)\n",
+                              (unsigned long long)cno);
+                       err = -EINVAL;
+               }
+               goto failed;
+       }
+
+       err = nilfs_ifile_read(sb, root, nilfs->ns_inode_size,
+                              &raw_cp->cp_ifile_inode, &root->ifile);
+       if (err)
+               goto failed_bh;
+
+       atomic64_set(&root->inodes_count,
+                       le64_to_cpu(raw_cp->cp_inodes_count));
+       atomic64_set(&root->blocks_count,
+                       le64_to_cpu(raw_cp->cp_blocks_count));
+
+       nilfs_cpfile_put_checkpoint(nilfs->ns_cpfile, cno, bh_cp);
+
+ reuse:
+       *rootp = root;
+       return 0;
+
+ failed_bh:
+       nilfs_cpfile_put_checkpoint(nilfs->ns_cpfile, cno, bh_cp);
+ failed:
+       nilfs_put_root(root);
+
+       return err;
+}
+
+static int nilfs_freeze(struct super_block *sb)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       int err;
+
+       if (sb->s_flags & MS_RDONLY)
+               return 0;
+
+       /* Mark super block clean */
+       down_write(&nilfs->ns_sem);
+       err = nilfs_cleanup_super(sb);
+       up_write(&nilfs->ns_sem);
+       return err;
+}
+
+static int nilfs_unfreeze(struct super_block *sb)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+
+       if (sb->s_flags & MS_RDONLY)
+               return 0;
+
+       down_write(&nilfs->ns_sem);
+       nilfs_setup_super(sb, false);
+       up_write(&nilfs->ns_sem);
+       return 0;
+}
+
+static int nilfs_statfs(struct dentry *dentry, struct kstatfs *buf)
+{
+       struct super_block *sb = dentry->d_sb;
+       struct nilfs_root *root = NILFS_I(d_inode(dentry))->i_root;
+       struct the_nilfs *nilfs = root->nilfs;
+       u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
+       unsigned long long blocks;
+       unsigned long overhead;
+       unsigned long nrsvblocks;
+       sector_t nfreeblocks;
+       u64 nmaxinodes, nfreeinodes;
+       int err;
+
+       /*
+        * Compute all of the segment blocks
+        *
+        * The blocks before first segment and after last segment
+        * are excluded.
+        */
+       blocks = nilfs->ns_blocks_per_segment * nilfs->ns_nsegments
+               - nilfs->ns_first_data_block;
+       nrsvblocks = nilfs->ns_nrsvsegs * nilfs->ns_blocks_per_segment;
+
+       /*
+        * Compute the overhead
+        *
+        * When distributing meta data blocks outside segment structure,
+        * We must count them as the overhead.
+        */
+       overhead = 0;
+
+       err = nilfs_count_free_blocks(nilfs, &nfreeblocks);
+       if (unlikely(err))
+               return err;
+
+       err = nilfs_ifile_count_free_inodes(root->ifile,
+                                           &nmaxinodes, &nfreeinodes);
+       if (unlikely(err)) {
+               printk(KERN_WARNING
+                       "NILFS warning: fail to count free inodes: err %d.\n",
+                       err);
+               if (err == -ERANGE) {
+                       /*
+                        * If nilfs_palloc_count_max_entries() returns
+                        * -ERANGE error code then we simply treat
+                        * curent inodes count as maximum possible and
+                        * zero as free inodes value.
+                        */
+                       nmaxinodes = atomic64_read(&root->inodes_count);
+                       nfreeinodes = 0;
+                       err = 0;
+               } else
+                       return err;
+       }
+
+       buf->f_type = NILFS_SUPER_MAGIC;
+       buf->f_bsize = sb->s_blocksize;
+       buf->f_blocks = blocks - overhead;
+       buf->f_bfree = nfreeblocks;
+       buf->f_bavail = (buf->f_bfree >= nrsvblocks) ?
+               (buf->f_bfree - nrsvblocks) : 0;
+       buf->f_files = nmaxinodes;
+       buf->f_ffree = nfreeinodes;
+       buf->f_namelen = NILFS_NAME_LEN;
+       buf->f_fsid.val[0] = (u32)id;
+       buf->f_fsid.val[1] = (u32)(id >> 32);
+
+       return 0;
+}
+
+static int nilfs_show_options(struct seq_file *seq, struct dentry *dentry)
+{
+       struct super_block *sb = dentry->d_sb;
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_root *root = NILFS_I(d_inode(dentry))->i_root;
+
+       if (!nilfs_test_opt(nilfs, BARRIER))
+               seq_puts(seq, ",nobarrier");
+       if (root->cno != NILFS_CPTREE_CURRENT_CNO)
+               seq_printf(seq, ",cp=%llu", (unsigned long long)root->cno);
+       if (nilfs_test_opt(nilfs, ERRORS_PANIC))
+               seq_puts(seq, ",errors=panic");
+       if (nilfs_test_opt(nilfs, ERRORS_CONT))
+               seq_puts(seq, ",errors=continue");
+       if (nilfs_test_opt(nilfs, STRICT_ORDER))
+               seq_puts(seq, ",order=strict");
+       if (nilfs_test_opt(nilfs, NORECOVERY))
+               seq_puts(seq, ",norecovery");
+       if (nilfs_test_opt(nilfs, DISCARD))
+               seq_puts(seq, ",discard");
+
+       return 0;
+}
+
+static const struct super_operations nilfs_sops = {
+       .alloc_inode    = nilfs_alloc_inode,
+       .destroy_inode  = nilfs_destroy_inode,
+       .dirty_inode    = nilfs_dirty_inode,
+       .evict_inode    = nilfs_evict_inode,
+       .put_super      = nilfs_put_super,
+       .sync_fs        = nilfs_sync_fs,
+       .freeze_fs      = nilfs_freeze,
+       .unfreeze_fs    = nilfs_unfreeze,
+       .statfs         = nilfs_statfs,
+       .remount_fs     = nilfs_remount,
+       .show_options = nilfs_show_options
+};
+
+enum {
+       Opt_err_cont, Opt_err_panic, Opt_err_ro,
+       Opt_barrier, Opt_nobarrier, Opt_snapshot, Opt_order, Opt_norecovery,
+       Opt_discard, Opt_nodiscard, Opt_err,
+};
+
+static match_table_t tokens = {
+       {Opt_err_cont, "errors=continue"},
+       {Opt_err_panic, "errors=panic"},
+       {Opt_err_ro, "errors=remount-ro"},
+       {Opt_barrier, "barrier"},
+       {Opt_nobarrier, "nobarrier"},
+       {Opt_snapshot, "cp=%u"},
+       {Opt_order, "order=%s"},
+       {Opt_norecovery, "norecovery"},
+       {Opt_discard, "discard"},
+       {Opt_nodiscard, "nodiscard"},
+       {Opt_err, NULL}
+};
+
+static int parse_options(char *options, struct super_block *sb, int is_remount)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       char *p;
+       substring_t args[MAX_OPT_ARGS];
+
+       if (!options)
+               return 1;
+
+       while ((p = strsep(&options, ",")) != NULL) {
+               int token;
+               if (!*p)
+                       continue;
+
+               token = match_token(p, tokens, args);
+               switch (token) {
+               case Opt_barrier:
+                       nilfs_set_opt(nilfs, BARRIER);
+                       break;
+               case Opt_nobarrier:
+                       nilfs_clear_opt(nilfs, BARRIER);
+                       break;
+               case Opt_order:
+                       if (strcmp(args[0].from, "relaxed") == 0)
+                               /* Ordered data semantics */
+                               nilfs_clear_opt(nilfs, STRICT_ORDER);
+                       else if (strcmp(args[0].from, "strict") == 0)
+                               /* Strict in-order semantics */
+                               nilfs_set_opt(nilfs, STRICT_ORDER);
+                       else
+                               return 0;
+                       break;
+               case Opt_err_panic:
+                       nilfs_write_opt(nilfs, ERROR_MODE, ERRORS_PANIC);
+                       break;
+               case Opt_err_ro:
+                       nilfs_write_opt(nilfs, ERROR_MODE, ERRORS_RO);
+                       break;
+               case Opt_err_cont:
+                       nilfs_write_opt(nilfs, ERROR_MODE, ERRORS_CONT);
+                       break;
+               case Opt_snapshot:
+                       if (is_remount) {
+                               printk(KERN_ERR
+                                      "NILFS: \"%s\" option is invalid "
+                                      "for remount.\n", p);
+                               return 0;
+                       }
+                       break;
+               case Opt_norecovery:
+                       nilfs_set_opt(nilfs, NORECOVERY);
+                       break;
+               case Opt_discard:
+                       nilfs_set_opt(nilfs, DISCARD);
+                       break;
+               case Opt_nodiscard:
+                       nilfs_clear_opt(nilfs, DISCARD);
+                       break;
+               default:
+                       printk(KERN_ERR
+                              "NILFS: Unrecognized mount option \"%s\"\n", p);
+                       return 0;
+               }
+       }
+       return 1;
+}
+
+static inline void
+nilfs_set_default_options(struct super_block *sb,
+                         struct nilfs_super_block *sbp)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+
+       nilfs->ns_mount_opt =
+               NILFS_MOUNT_ERRORS_RO | NILFS_MOUNT_BARRIER;
+}
+
+static int nilfs_setup_super(struct super_block *sb, int is_mount)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_super_block **sbp;
+       int max_mnt_count;
+       int mnt_count;
+
+       /* nilfs->ns_sem must be locked by the caller. */
+       sbp = nilfs_prepare_super(sb, 0);
+       if (!sbp)
+               return -EIO;
+
+       if (!is_mount)
+               goto skip_mount_setup;
+
+       max_mnt_count = le16_to_cpu(sbp[0]->s_max_mnt_count);
+       mnt_count = le16_to_cpu(sbp[0]->s_mnt_count);
+
+       if (nilfs->ns_mount_state & NILFS_ERROR_FS) {
+               printk(KERN_WARNING
+                      "NILFS warning: mounting fs with errors\n");
+#if 0
+       } else if (max_mnt_count >= 0 && mnt_count >= max_mnt_count) {
+               printk(KERN_WARNING
+                      "NILFS warning: maximal mount count reached\n");
+#endif
+       }
+       if (!max_mnt_count)
+               sbp[0]->s_max_mnt_count = cpu_to_le16(NILFS_DFL_MAX_MNT_COUNT);
+
+       sbp[0]->s_mnt_count = cpu_to_le16(mnt_count + 1);
+       sbp[0]->s_mtime = cpu_to_le64(get_seconds());
+
+skip_mount_setup:
+       sbp[0]->s_state =
+               cpu_to_le16(le16_to_cpu(sbp[0]->s_state) & ~NILFS_VALID_FS);
+       /* synchronize sbp[1] with sbp[0] */
+       if (sbp[1])
+               memcpy(sbp[1], sbp[0], nilfs->ns_sbsize);
+       return nilfs_commit_super(sb, NILFS_SB_COMMIT_ALL);
+}
+
+struct nilfs_super_block *nilfs_read_super_block(struct super_block *sb,
+                                                u64 pos, int blocksize,
+                                                struct buffer_head **pbh)
+{
+       unsigned long long sb_index = pos;
+       unsigned long offset;
+
+       offset = do_div(sb_index, blocksize);
+       *pbh = sb_bread(sb, sb_index);
+       if (!*pbh)
+               return NULL;
+       return (struct nilfs_super_block *)((char *)(*pbh)->b_data + offset);
+}
+
+int nilfs_store_magic_and_option(struct super_block *sb,
+                                struct nilfs_super_block *sbp,
+                                char *data)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+
+       sb->s_magic = le16_to_cpu(sbp->s_magic);
+
+       /* FS independent flags */
+#ifdef NILFS_ATIME_DISABLE
+       sb->s_flags |= MS_NOATIME;
+#endif
+
+       nilfs_set_default_options(sb, sbp);
+
+       nilfs->ns_resuid = le16_to_cpu(sbp->s_def_resuid);
+       nilfs->ns_resgid = le16_to_cpu(sbp->s_def_resgid);
+       nilfs->ns_interval = le32_to_cpu(sbp->s_c_interval);
+       nilfs->ns_watermark = le32_to_cpu(sbp->s_c_block_max);
+
+       return !parse_options(data, sb, 0) ? -EINVAL : 0 ;
+}
+
+int nilfs_check_feature_compatibility(struct super_block *sb,
+                                     struct nilfs_super_block *sbp)
+{
+       __u64 features;
+
+       features = le64_to_cpu(sbp->s_feature_incompat) &
+               ~NILFS_FEATURE_INCOMPAT_SUPP;
+       if (features) {
+               printk(KERN_ERR "NILFS: couldn't mount because of unsupported "
+                      "optional features (%llx)\n",
+                      (unsigned long long)features);
+               return -EINVAL;
+       }
+       features = le64_to_cpu(sbp->s_feature_compat_ro) &
+               ~NILFS_FEATURE_COMPAT_RO_SUPP;
+       if (!(sb->s_flags & MS_RDONLY) && features) {
+               printk(KERN_ERR "NILFS: couldn't mount RDWR because of "
+                      "unsupported optional features (%llx)\n",
+                      (unsigned long long)features);
+               return -EINVAL;
+       }
+       return 0;
+}
+
+static int nilfs_get_root_dentry(struct super_block *sb,
+                                struct nilfs_root *root,
+                                struct dentry **root_dentry)
+{
+       struct inode *inode;
+       struct dentry *dentry;
+       int ret = 0;
+
+       inode = nilfs_iget(sb, root, NILFS_ROOT_INO);
+       if (IS_ERR(inode)) {
+               printk(KERN_ERR "NILFS: get root inode failed\n");
+               ret = PTR_ERR(inode);
+               goto out;
+       }
+       if (!S_ISDIR(inode->i_mode) || !inode->i_blocks || !inode->i_size) {
+               iput(inode);
+               printk(KERN_ERR "NILFS: corrupt root inode.\n");
+               ret = -EINVAL;
+               goto out;
+       }
+
+       if (root->cno == NILFS_CPTREE_CURRENT_CNO) {
+               dentry = d_find_alias(inode);
+               if (!dentry) {
+                       dentry = d_make_root(inode);
+                       if (!dentry) {
+                               ret = -ENOMEM;
+                               goto failed_dentry;
+                       }
+               } else {
+                       iput(inode);
+               }
+       } else {
+               dentry = d_obtain_root(inode);
+               if (IS_ERR(dentry)) {
+                       ret = PTR_ERR(dentry);
+                       goto failed_dentry;
+               }
+       }
+       *root_dentry = dentry;
+ out:
+       return ret;
+
+ failed_dentry:
+       printk(KERN_ERR "NILFS: get root dentry failed\n");
+       goto out;
+}
+
+static int nilfs_attach_snapshot(struct super_block *s, __u64 cno,
+                                struct dentry **root_dentry)
+{
+       struct the_nilfs *nilfs = s->s_fs_info;
+       struct nilfs_root *root;
+       int ret;
+
+       mutex_lock(&nilfs->ns_snapshot_mount_mutex);
+
+       down_read(&nilfs->ns_segctor_sem);
+       ret = nilfs_cpfile_is_snapshot(nilfs->ns_cpfile, cno);
+       up_read(&nilfs->ns_segctor_sem);
+       if (ret < 0) {
+               ret = (ret == -ENOENT) ? -EINVAL : ret;
+               goto out;
+       } else if (!ret) {
+               printk(KERN_ERR "NILFS: The specified checkpoint is "
+                      "not a snapshot (checkpoint number=%llu).\n",
+                      (unsigned long long)cno);
+               ret = -EINVAL;
+               goto out;
+       }
+
+       ret = nilfs_attach_checkpoint(s, cno, false, &root);
+       if (ret) {
+               printk(KERN_ERR "NILFS: error loading snapshot "
+                      "(checkpoint number=%llu).\n",
+              (unsigned long long)cno);
+               goto out;
+       }
+       ret = nilfs_get_root_dentry(s, root, root_dentry);
+       nilfs_put_root(root);
+ out:
+       mutex_unlock(&nilfs->ns_snapshot_mount_mutex);
+       return ret;
+}
+
+/**
+ * nilfs_tree_is_busy() - try to shrink dentries of a checkpoint
+ * @root_dentry: root dentry of the tree to be shrunk
+ *
+ * This function returns true if the tree was in-use.
+ */
+static bool nilfs_tree_is_busy(struct dentry *root_dentry)
+{
+       shrink_dcache_parent(root_dentry);
+       return d_count(root_dentry) > 1;
+}
+
+int nilfs_checkpoint_is_mounted(struct super_block *sb, __u64 cno)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       struct nilfs_root *root;
+       struct inode *inode;
+       struct dentry *dentry;
+       int ret;
+
+       if (cno > nilfs->ns_cno)
+               return false;
+
+       if (cno >= nilfs_last_cno(nilfs))
+               return true;    /* protect recent checkpoints */
+
+       ret = false;
+       root = nilfs_lookup_root(nilfs, cno);
+       if (root) {
+               inode = nilfs_ilookup(sb, root, NILFS_ROOT_INO);
+               if (inode) {
+                       dentry = d_find_alias(inode);
+                       if (dentry) {
+                               ret = nilfs_tree_is_busy(dentry);
+                               dput(dentry);
+                       }
+                       iput(inode);
+               }
+               nilfs_put_root(root);
+       }
+       return ret;
+}
+
+/**
+ * nilfs_fill_super() - initialize a super block instance
+ * @sb: super_block
+ * @data: mount options
+ * @silent: silent mode flag
+ *
+ * This function is called exclusively by nilfs->ns_mount_mutex.
+ * So, the recovery process is protected from other simultaneous mounts.
+ */
+static int
+nilfs_fill_super(struct super_block *sb, void *data, int silent)
+{
+       struct the_nilfs *nilfs;
+       struct nilfs_root *fsroot;
+       __u64 cno;
+       int err;
+
+       nilfs = alloc_nilfs(sb->s_bdev);
+       if (!nilfs)
+               return -ENOMEM;
+
+       sb->s_fs_info = nilfs;
+
+       err = init_nilfs(nilfs, sb, (char *)data);
+       if (err)
+               goto failed_nilfs;
+
+       sb->s_op = &nilfs_sops;
+       sb->s_export_op = &nilfs_export_ops;
+       sb->s_root = NULL;
+       sb->s_time_gran = 1;
+       sb->s_max_links = NILFS_LINK_MAX;
+
+       sb->s_bdi = &bdev_get_queue(sb->s_bdev)->backing_dev_info;
+
+       err = load_nilfs(nilfs, sb);
+       if (err)
+               goto failed_nilfs;
+
+       cno = nilfs_last_cno(nilfs);
+       err = nilfs_attach_checkpoint(sb, cno, true, &fsroot);
+       if (err) {
+               printk(KERN_ERR "NILFS: error loading last checkpoint "
+                      "(checkpoint number=%llu).\n", (unsigned long long)cno);
+               goto failed_unload;
+       }
+
+       if (!(sb->s_flags & MS_RDONLY)) {
+               err = nilfs_attach_log_writer(sb, fsroot);
+               if (err)
+                       goto failed_checkpoint;
+       }
+
+       err = nilfs_get_root_dentry(sb, fsroot, &sb->s_root);
+       if (err)
+               goto failed_segctor;
+
+       nilfs_put_root(fsroot);
+
+       if (!(sb->s_flags & MS_RDONLY)) {
+               down_write(&nilfs->ns_sem);
+               nilfs_setup_super(sb, true);
+               up_write(&nilfs->ns_sem);
+       }
+
+       return 0;
+
+ failed_segctor:
+       nilfs_detach_log_writer(sb);
+
+ failed_checkpoint:
+       nilfs_put_root(fsroot);
+
+ failed_unload:
+       iput(nilfs->ns_sufile);
+       iput(nilfs->ns_cpfile);
+       iput(nilfs->ns_dat);
+
+ failed_nilfs:
+       destroy_nilfs(nilfs);
+       return err;
+}
+
+static int nilfs_remount(struct super_block *sb, int *flags, char *data)
+{
+       struct the_nilfs *nilfs = sb->s_fs_info;
+       unsigned long old_sb_flags;
+       unsigned long old_mount_opt;
+       int err;
+
+       sync_filesystem(sb);
+       old_sb_flags = sb->s_flags;
+       old_mount_opt = nilfs->ns_mount_opt;
+
+       if (!parse_options(data, sb, 1)) {
+               err = -EINVAL;
+               goto restore_opts;
+       }
+       sb->s_flags = (sb->s_flags & ~MS_POSIXACL);
+
+       err = -EINVAL;
+
+       if (!nilfs_valid_fs(nilfs)) {
+               printk(KERN_WARNING "NILFS (device %s): couldn't "
+                      "remount because the filesystem is in an "
+                      "incomplete recovery state.\n", sb->s_id);
+               goto restore_opts;
+       }
+
+       if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
+               goto out;
+       if (*flags & MS_RDONLY) {
+               /* Shutting down log writer */
+               nilfs_detach_log_writer(sb);
+               sb->s_flags |= MS_RDONLY;
+
+               /*
+                * Remounting a valid RW partition RDONLY, so set
+                * the RDONLY flag and then mark the partition as valid again.
+                */
+               down_write(&nilfs->ns_sem);
+               nilfs_cleanup_super(sb);
+               up_write(&nilfs->ns_sem);
+       } else {
+               __u64 features;
+               struct nilfs_root *root;
+
+               /*
+                * Mounting a RDONLY partition read-write, so reread and
+                * store the current valid flag.  (It may have been changed
+                * by fsck since we originally mounted the partition.)
+                */
+               down_read(&nilfs->ns_sem);
+               features = le64_to_cpu(nilfs->ns_sbp[0]->s_feature_compat_ro) &
+                       ~NILFS_FEATURE_COMPAT_RO_SUPP;
+               up_read(&nilfs->ns_sem);
+               if (features) {
+                       printk(KERN_WARNING "NILFS (device %s): couldn't "
+                              "remount RDWR because of unsupported optional "
+                              "features (%llx)\n",
+                              sb->s_id, (unsigned long long)features);
+                       err = -EROFS;
+                       goto restore_opts;
+               }
+
+               sb->s_flags &= ~MS_RDONLY;
+
+               root = NILFS_I(d_inode(sb->s_root))->i_root;
+               err = nilfs_attach_log_writer(sb, root);
+               if (err)
+                       goto restore_opts;
+
+               down_write(&nilfs->ns_sem);
+               nilfs_setup_super(sb, true);
+               up_write(&nilfs->ns_sem);
+       }
+ out:
+       return 0;
+
+ restore_opts:
+       sb->s_flags = old_sb_flags;
+       nilfs->ns_mount_opt = old_mount_opt;
+       return err;
+}
+
+struct nilfs_super_data {
+       struct block_device *bdev;
+       __u64 cno;
+       int flags;
+};
+
+/**
+ * nilfs_identify - pre-read mount options needed to identify mount instance
+ * @data: mount options
+ * @sd: nilfs_super_data
+ */
+static int nilfs_identify(char *data, struct nilfs_super_data *sd)
+{
+       char *p, *options = data;
+       substring_t args[MAX_OPT_ARGS];
+       int token;
+       int ret = 0;
+
+       do {
+               p = strsep(&options, ",");
+               if (p != NULL && *p) {
+                       token = match_token(p, tokens, args);
+                       if (token == Opt_snapshot) {
+                               if (!(sd->flags & MS_RDONLY)) {
+                                       ret++;
+                               } else {
+                                       sd->cno = simple_strtoull(args[0].from,
+                                                                 NULL, 0);
+                                       /*
+                                        * No need to see the end pointer;
+                                        * match_token() has done syntax
+                                        * checking.
+                                        */
+                                       if (sd->cno == 0)
+                                               ret++;
+                               }
+                       }
+                       if (ret)
+                               printk(KERN_ERR
+                                      "NILFS: invalid mount option: %s\n", p);
+               }
+               if (!options)
+                       break;
+               BUG_ON(options == data);
+               *(options - 1) = ',';
+       } while (!ret);
+       return ret;
+}
+
+static int nilfs_set_bdev_super(struct super_block *s, void *data)
+{
+       s->s_bdev = data;
+       s->s_dev = s->s_bdev->bd_dev;
+       return 0;
+}
+
+static int nilfs_test_bdev_super(struct super_block *s, void *data)
+{
+       return (void *)s->s_bdev == data;
+}
+
+static struct dentry *
+nilfs_mount(struct file_system_type *fs_type, int flags,
+            const char *dev_name, void *data)
+{
+       struct nilfs_super_data sd;
+       struct super_block *s;
+       fmode_t mode = FMODE_READ | FMODE_EXCL;
+       struct dentry *root_dentry;
+       int err, s_new = false;
+
+       if (!(flags & MS_RDONLY))
+               mode |= FMODE_WRITE;
+
+       sd.bdev = blkdev_get_by_path(dev_name, mode, fs_type);
+       if (IS_ERR(sd.bdev))
+               return ERR_CAST(sd.bdev);
+
+       sd.cno = 0;
+       sd.flags = flags;
+       if (nilfs_identify((char *)data, &sd)) {
+               err = -EINVAL;
+               goto failed;
+       }
+
+       /*
+        * once the super is inserted into the list by sget, s_umount
+        * will protect the lockfs code from trying to start a snapshot
+        * while we are mounting
+        */
+       mutex_lock(&sd.bdev->bd_fsfreeze_mutex);
+       if (sd.bdev->bd_fsfreeze_count > 0) {
+               mutex_unlock(&sd.bdev->bd_fsfreeze_mutex);
+               err = -EBUSY;
+               goto failed;
+       }
+       s = sget(fs_type, nilfs_test_bdev_super, nilfs_set_bdev_super, flags,
+                sd.bdev);
+       mutex_unlock(&sd.bdev->bd_fsfreeze_mutex);
+       if (IS_ERR(s)) {
+               err = PTR_ERR(s);
+               goto failed;
+       }
+
+       if (!s->s_root) {
+               char b[BDEVNAME_SIZE];
+
+               s_new = true;
+
+               /* New superblock instance created */
+               s->s_mode = mode;
+               strlcpy(s->s_id, bdevname(sd.bdev, b), sizeof(s->s_id));
+               sb_set_blocksize(s, block_size(sd.bdev));
+
+               err = nilfs_fill_super(s, data, flags & MS_SILENT ? 1 : 0);
+               if (err)
+                       goto failed_super;
+
+               s->s_flags |= MS_ACTIVE;
+       } else if (!sd.cno) {
+               if (nilfs_tree_is_busy(s->s_root)) {
+                       if ((flags ^ s->s_flags) & MS_RDONLY) {
+                               printk(KERN_ERR "NILFS: the device already "
+                                      "has a %s mount.\n",
+                                      (s->s_flags & MS_RDONLY) ?
+                                      "read-only" : "read/write");
+                               err = -EBUSY;
+                               goto failed_super;
+                       }
+               } else {
+                       /*
+                        * Try remount to setup mount states if the current
+                        * tree is not mounted and only snapshots use this sb.
+                        */
+                       err = nilfs_remount(s, &flags, data);
+                       if (err)
+                               goto failed_super;
+               }
+       }
+
+       if (sd.cno) {
+               err = nilfs_attach_snapshot(s, sd.cno, &root_dentry);
+               if (err)
+                       goto failed_super;
+       } else {
+               root_dentry = dget(s->s_root);
+       }
+
+       if (!s_new)
+               blkdev_put(sd.bdev, mode);
+
+       return root_dentry;
+
+ failed_super:
+       deactivate_locked_super(s);
+
+ failed:
+       if (!s_new)
+               blkdev_put(sd.bdev, mode);
+       return ERR_PTR(err);
+}
+
+struct file_system_type nilfs_fs_type = {
+       .owner    = THIS_MODULE,
+       .name     = "nilfs2",
+       .mount    = nilfs_mount,
+       .kill_sb  = kill_block_super,
+       .fs_flags = FS_REQUIRES_DEV,
+};
+MODULE_ALIAS_FS("nilfs2");
+
+static void nilfs_inode_init_once(void *obj)
+{
+       struct nilfs_inode_info *ii = obj;
+
+       INIT_LIST_HEAD(&ii->i_dirty);
+#ifdef CONFIG_NILFS_XATTR
+       init_rwsem(&ii->xattr_sem);
+#endif
+       address_space_init_once(&ii->i_btnode_cache);
+       ii->i_bmap = &ii->i_bmap_data;
+       inode_init_once(&ii->vfs_inode);
+}
+
+static void nilfs_segbuf_init_once(void *obj)
+{
+       memset(obj, 0, sizeof(struct nilfs_segment_buffer));
+}
+
+static void nilfs_destroy_cachep(void)
+{
+       /*
+        * Make sure all delayed rcu free inodes are flushed before we
+        * destroy cache.
+        */
+       rcu_barrier();
+
+       if (nilfs_inode_cachep)
+               kmem_cache_destroy(nilfs_inode_cachep);
+       if (nilfs_transaction_cachep)
+               kmem_cache_destroy(nilfs_transaction_cachep);
+       if (nilfs_segbuf_cachep)
+               kmem_cache_destroy(nilfs_segbuf_cachep);
+       if (nilfs_btree_path_cache)
+               kmem_cache_destroy(nilfs_btree_path_cache);
+}
+
+static int __init nilfs_init_cachep(void)
+{
+       nilfs_inode_cachep = kmem_cache_create("nilfs2_inode_cache",
+                       sizeof(struct nilfs_inode_info), 0,
+                       SLAB_RECLAIM_ACCOUNT, nilfs_inode_init_once);
+       if (!nilfs_inode_cachep)
+               goto fail;
+
+       nilfs_transaction_cachep = kmem_cache_create("nilfs2_transaction_cache",
+                       sizeof(struct nilfs_transaction_info), 0,
+                       SLAB_RECLAIM_ACCOUNT, NULL);
+       if (!nilfs_transaction_cachep)
+               goto fail;
+
+       nilfs_segbuf_cachep = kmem_cache_create("nilfs2_segbuf_cache",
+                       sizeof(struct nilfs_segment_buffer), 0,
+                       SLAB_RECLAIM_ACCOUNT, nilfs_segbuf_init_once);
+       if (!nilfs_segbuf_cachep)
+               goto fail;
+
+       nilfs_btree_path_cache = kmem_cache_create("nilfs2_btree_path_cache",
+                       sizeof(struct nilfs_btree_path) * NILFS_BTREE_LEVEL_MAX,
+                       0, 0, NULL);
+       if (!nilfs_btree_path_cache)
+               goto fail;
+
+       return 0;
+
+fail:
+       nilfs_destroy_cachep();
+       return -ENOMEM;
+}
+
+static int __init init_nilfs_fs(void)
+{
+       int err;
+
+       err = nilfs_init_cachep();
+       if (err)
+               goto fail;
+
+       err = nilfs_sysfs_init();
+       if (err)
+               goto free_cachep;
+
+       err = register_filesystem(&nilfs_fs_type);
+       if (err)
+               goto deinit_sysfs_entry;
+
+       printk(KERN_INFO "NILFS version 2 loaded\n");
+       return 0;
+
+deinit_sysfs_entry:
+       nilfs_sysfs_exit();
+free_cachep:
+       nilfs_destroy_cachep();
+fail:
+       return err;
+}
+
+static void __exit exit_nilfs_fs(void)
+{
+       nilfs_destroy_cachep();
+       nilfs_sysfs_exit();
+       unregister_filesystem(&nilfs_fs_type);
+}
+
+module_init(init_nilfs_fs)
+module_exit(exit_nilfs_fs)