x86/kprobes: Fix crash when probing CS CALL instructionsgrafted

When using eBPF to probe CS CALL instructions within a function,
a crash can be triggered.

The eBPF tool probes offset 257 of the __hrtimer_run_queues()
function:

<__hrtimer_run_queues+249>:  nopl   0x0(%rax,%rax,1)
<__hrtimer_run_queues+254>:  mov    %r14,%rdi
<__hrtimer_run_queues+257>:  cs call <__x86_indirect_thunk_r12>
<__hrtimer_run_queues+263>:  mov    %eax,%r12d
<__hrtimer_run_queues+266>:  xchg   %ax,%ax
<__hrtimer_run_queues+268>:  mov    %r13,%rdi

Which triggers this crash:

  BUG: unable to handle page fault for address: 00000000000f41c9
  #PF: supervisor write access in kernel mode
  #PF: error_code(0x0002) - not-present page
  PGD 0 P4D 0
  Oops: 0002 [#1] SMP NOPTI
  CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P
  RIP: 0010:__hrtimer_run_queues+0x106/0x230

Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is
at the 6th byte of the above CS CALL instruction. Since the CS CALL
instruction occupies 6 bytes, the exception occurred in the middle of that
call instruction.

The root cause is that when using eBPF tools to probe in the middle of a
function, a kprobe with INT3 is used as the underlying implementation.

During single-step emulation of the original CALL instruction,
int3_emulate_call() assumes that the probed CALL instruction is 5 bytes
long. However, the actual CS-prefixed CALL instruction occupies 6 bytes,
so it constructs an incorrect exception return address. When the CPU
returns from the kprobe handler, the next instruction to be executed is at
the address of the last byte of that CS CALL instruction. Coincidentally,
starting from that address, the CPU fetches and decodes a completely
different instruction, which ultimately triggers a kernel crash.

Fix the issue by using the actual instruction length obtained from
the instruction decoder when constructing the exception return
address, rather than relying on the hardcoded CALL_INSN_SIZE macro.

[ mingo: Refined the changelog ]

Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step")
Suggested-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Jinke Han <jinkehan@didiglobal.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000

1 files changed, 1413 insertions, 0 deletions

diff --git a/fs/nilfs2/super.c b/fs/nilfs2/super.c
new file mode 100644
index 000000000..893a504cb
--- /dev/null
+++ b/fs/nilfs2/super.c
@@ -0,0 +1,1413 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS module and super block management.
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Ryusuke Konishi.
+ */
+/*
+ *  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/crc32.h>
+#include <linux/vfs.h>
+#include <linux/writeback.h>
+#include <linux/seq_file.h>
+#include <linux/mount.h>
+#include <linux/fs_context.h>
+#include <linux/fs_parser.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);
+
+void __nilfs_msg(struct super_block *sb, const char *fmt, ...)
+{
+	struct va_format vaf;
+	va_list args;
+	int level;
+
+	va_start(args, fmt);
+
+	level = printk_get_level(fmt);
+	vaf.fmt = printk_skip_level(fmt);
+	vaf.va = &args;
+
+	if (sb)
+		printk("%c%cNILFS (%s): %pV\n",
+		       KERN_SOH_ASCII, level, sb->s_id, &vaf);
+	else
+		printk("%c%cNILFS: %pV\n",
+		       KERN_SOH_ASCII, level, &vaf);
+
+	va_end(args);
+}
+
+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
+ * @sb:       super block instance
+ * @function: name of calling function
+ * @fmt:      format string for message to be output
+ * @...:      optional arguments to @fmt
+ *
+ * __nilfs_error() sets an ERROR_FS flag on the superblock as well as
+ * reporting an error message.  This function should be called when
+ * NILFS detects incoherences or defects of meta data on disk.
+ *
+ * This implements the body of nilfs_error() macro.  Normally,
+ * nilfs_error() should be used.  As for sustainable errors such as a
+ * single-shot I/O error, nilfs_err() should be used instead.
+ *
+ * Callers should not add a trailing newline since this will do it.
+ */
+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_rdonly(sb)) {
+		nilfs_set_error(sb);
+
+		if (nilfs_test_opt(nilfs, ERRORS_RO)) {
+			printk(KERN_CRIT "Remounting filesystem read-only\n");
+			sb->s_flags |= SB_RDONLY;
+		}
+	}
+
+	if (nilfs_test_opt(nilfs, ERRORS_PANIC))
+		panic("NILFS (device %s): panic forced after error\n",
+		      sb->s_id);
+}
+
+struct inode *nilfs_alloc_inode(struct super_block *sb)
+{
+	struct nilfs_inode_info *ii;
+
+	ii = alloc_inode_sb(sb, nilfs_inode_cachep, GFP_NOFS);
+	if (!ii)
+		return NULL;
+	ii->i_bh = NULL;
+	ii->i_state = 0;
+	ii->i_type = 0;
+	ii->i_cno = 0;
+	ii->i_assoc_inode = NULL;
+	ii->i_bmap = &ii->i_bmap_data;
+	return &ii->vfs_inode;
+}
+
+static void nilfs_free_inode(struct inode *inode)
+{
+	if (nilfs_is_metadata_file_inode(inode))
+		nilfs_mdt_destroy(inode);
+
+	kmem_cache_free(nilfs_inode_cachep, NILFS_I(inode));
+}
+
+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],
+					  REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
+	} else {
+		err = sync_dirty_buffer(nilfs->ns_sbh[0]);
+	}
+
+	if (unlikely(err)) {
+		nilfs_err(sb, "unable to write superblock: err=%d", 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 {
+			nilfs_crit(sb, "superblock broke");
+			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;
+	time64_t t;
+
+	/* nilfs->ns_sem must be locked by the caller. */
+	t = ktime_get_real_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.
+ *
+ * Return: 0 on success, %-EIO if I/O error or superblock is corrupted.
+ */
+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)
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+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;  /* 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;
+	} else {
+		sb2i = -1;
+		blocknr = 0;
+	}
+	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) {
+		nilfs_warn(sb,
+			   "unable to move secondary superblock to block %llu",
+			   (unsigned long long)newblocknr);
+		ret = -EIO;
+		goto out;
+	}
+	nsbp = (void *)nsbh->b_data + offset;
+
+	lock_buffer(nsbh);
+	if (sb2i >= 0) {
+		/*
+		 * The position of the second superblock only changes by 4KiB,
+		 * which is larger than the maximum superblock data size
+		 * (= 1KiB), so there is no need to use memmove() to allow
+		 * overlap between source and destination.
+		 */
+		memcpy(nsbp, nilfs->ns_sbp[sb2i], nilfs->ns_sbsize);
+
+		/*
+		 * Zero fill after copy to avoid overwriting in case of move
+		 * within the same block.
+		 */
+		memset(nsbh->b_data, 0, offset);
+		memset((void *)nsbp + nilfs->ns_sbsize, 0,
+		       nsbh->b_size - offset - nilfs->ns_sbsize);
+	} else {
+		memset(nsbh->b_data, 0, nsbh->b_size);
+	}
+	set_buffer_uptodate(nsbh);
+	unlock_buffer(nsbh);
+
+	if (sb2i >= 0) {
+		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)
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+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 = bdev_nr_bytes(sb->s_bdev);
+	if (newsize > devsize)
+		goto out;
+
+	/*
+	 * Prevent underflow in second superblock position calculation.
+	 * The exact minimum size check is done in nilfs_sufile_resize().
+	 */
+	if (newsize < 4096) {
+		ret = -ENOSPC;
+		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;
+	newnsegs = div64_ul(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_rdonly(sb)) {
+		down_write(&nilfs->ns_sem);
+		nilfs_cleanup_super(sb);
+		up_write(&nilfs->ns_sem);
+	}
+
+	nilfs_sysfs_delete_device_group(nilfs);
+	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;
+	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_ifile_read(sb, root, cno, nilfs->ns_inode_size);
+	up_read(&nilfs->ns_segctor_sem);
+	if (unlikely(err))
+		goto failed;
+
+ reuse:
+	*rootp = root;
+	return 0;
+
+ failed:
+	if (err == -EINVAL)
+		nilfs_err(sb, "Invalid checkpoint (checkpoint number=%llu)",
+			  (unsigned long long)cno);
+	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_rdonly(sb))
+		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_rdonly(sb))
+		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)) {
+		nilfs_warn(sb, "failed to count free inodes: err=%d", 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 = u64_to_fsid(id);
+
+	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,
+	.free_inode     = nilfs_free_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,
+	.show_options = nilfs_show_options
+};
+
+enum {
+	Opt_err, Opt_barrier, Opt_snapshot, Opt_order, Opt_norecovery,
+	Opt_discard,
+};
+
+static const struct constant_table nilfs_param_err[] = {
+	{"continue",	NILFS_MOUNT_ERRORS_CONT},
+	{"panic",	NILFS_MOUNT_ERRORS_PANIC},
+	{"remount-ro",	NILFS_MOUNT_ERRORS_RO},
+	{}
+};
+
+static const struct fs_parameter_spec nilfs_param_spec[] = {
+	fsparam_enum	("errors", Opt_err, nilfs_param_err),
+	fsparam_flag_no	("barrier", Opt_barrier),
+	fsparam_u64	("cp", Opt_snapshot),
+	fsparam_string	("order", Opt_order),
+	fsparam_flag	("norecovery", Opt_norecovery),
+	fsparam_flag_no	("discard", Opt_discard),
+	{}
+};
+
+struct nilfs_fs_context {
+	unsigned long ns_mount_opt;
+	__u64 cno;
+};
+
+static int nilfs_parse_param(struct fs_context *fc, struct fs_parameter *param)
+{
+	struct nilfs_fs_context *nilfs = fc->fs_private;
+	int is_remount = fc->purpose == FS_CONTEXT_FOR_RECONFIGURE;
+	struct fs_parse_result result;
+	int opt;
+
+	opt = fs_parse(fc, nilfs_param_spec, param, &result);
+	if (opt < 0)
+		return opt;
+
+	switch (opt) {
+	case Opt_barrier:
+		if (result.negated)
+			nilfs_clear_opt(nilfs, BARRIER);
+		else
+			nilfs_set_opt(nilfs, BARRIER);
+		break;
+	case Opt_order:
+		if (strcmp(param->string, "relaxed") == 0)
+			/* Ordered data semantics */
+			nilfs_clear_opt(nilfs, STRICT_ORDER);
+		else if (strcmp(param->string, "strict") == 0)
+			/* Strict in-order semantics */
+			nilfs_set_opt(nilfs, STRICT_ORDER);
+		else
+			return -EINVAL;
+		break;
+	case Opt_err:
+		nilfs->ns_mount_opt &= ~NILFS_MOUNT_ERROR_MODE;
+		nilfs->ns_mount_opt |= result.uint_32;
+		break;
+	case Opt_snapshot:
+		if (is_remount) {
+			struct super_block *sb = fc->root->d_sb;
+
+			nilfs_err(sb,
+				  "\"%s\" option is invalid for remount",
+				  param->key);
+			return -EINVAL;
+		}
+		if (result.uint_64 == 0) {
+			nilfs_err(NULL,
+				  "invalid option \"cp=0\": invalid checkpoint number 0");
+			return -EINVAL;
+		}
+		nilfs->cno = result.uint_64;
+		break;
+	case Opt_norecovery:
+		nilfs_set_opt(nilfs, NORECOVERY);
+		break;
+	case Opt_discard:
+		if (result.negated)
+			nilfs_clear_opt(nilfs, DISCARD);
+		else
+			nilfs_set_opt(nilfs, DISCARD);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+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) {
+		nilfs_warn(sb, "mounting fs with errors");
+#if 0
+	} else if (max_mnt_count >= 0 && mnt_count >= max_mnt_count) {
+		nilfs_warn(sb, "maximal mount count reached");
+#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(ktime_get_real_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(struct super_block *sb,
+		      struct nilfs_super_block *sbp)
+{
+	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 |= SB_NOATIME;
+#endif
+
+	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 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) {
+		nilfs_err(sb,
+			  "couldn't mount because of unsupported optional features (%llx)",
+			  (unsigned long long)features);
+		return -EINVAL;
+	}
+	features = le64_to_cpu(sbp->s_feature_compat_ro) &
+		~NILFS_FEATURE_COMPAT_RO_SUPP;
+	if (!sb_rdonly(sb) && features) {
+		nilfs_err(sb,
+			  "couldn't mount RDWR because of unsupported optional features (%llx)",
+			  (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)) {
+		ret = PTR_ERR(inode);
+		nilfs_err(sb, "error %d getting root inode", ret);
+		goto out;
+	}
+	if (!S_ISDIR(inode->i_mode) || !inode->i_blocks || !inode->i_size) {
+		iput(inode);
+		nilfs_err(sb, "corrupt root inode");
+		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:
+	nilfs_err(sb, "error %d getting root dentry", ret);
+	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) {
+		nilfs_err(s,
+			  "The specified checkpoint is not a snapshot (checkpoint number=%llu)",
+			  (unsigned long long)cno);
+		ret = -EINVAL;
+		goto out;
+	}
+
+	ret = nilfs_attach_checkpoint(s, cno, false, &root);
+	if (ret) {
+		nilfs_err(s,
+			  "error %d while loading snapshot (checkpoint number=%llu)",
+			  ret, (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
+ *
+ * Return: true if the tree was in-use, false otherwise.
+ */
+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
+ * @fc: filesystem context
+ *
+ * This function is called exclusively by nilfs->ns_mount_mutex.
+ * So, the recovery process is protected from other simultaneous mounts.
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+static int
+nilfs_fill_super(struct super_block *sb, struct fs_context *fc)
+{
+	struct the_nilfs *nilfs;
+	struct nilfs_root *fsroot;
+	struct nilfs_fs_context *ctx = fc->fs_private;
+	__u64 cno;
+	int err;
+
+	nilfs = alloc_nilfs(sb);
+	if (!nilfs)
+		return -ENOMEM;
+
+	sb->s_fs_info = nilfs;
+
+	err = init_nilfs(nilfs, sb);
+	if (err)
+		goto failed_nilfs;
+
+	/* Copy in parsed mount options */
+	nilfs->ns_mount_opt = ctx->ns_mount_opt;
+
+	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;
+
+	err = load_nilfs(nilfs, sb);
+	if (err)
+		goto failed_nilfs;
+
+	super_set_uuid(sb, nilfs->ns_sbp[0]->s_uuid,
+		       sizeof(nilfs->ns_sbp[0]->s_uuid));
+	super_set_sysfs_name_bdev(sb);
+
+	cno = nilfs_last_cno(nilfs);
+	err = nilfs_attach_checkpoint(sb, cno, true, &fsroot);
+	if (err) {
+		nilfs_err(sb,
+			  "error %d while loading last checkpoint (checkpoint number=%llu)",
+			  err, (unsigned long long)cno);
+		goto failed_unload;
+	}
+
+	if (!sb_rdonly(sb)) {
+		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_rdonly(sb)) {
+		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:
+	nilfs_sysfs_delete_device_group(nilfs);
+	iput(nilfs->ns_sufile);
+	iput(nilfs->ns_cpfile);
+	iput(nilfs->ns_dat);
+
+ failed_nilfs:
+	destroy_nilfs(nilfs);
+	return err;
+}
+
+static int nilfs_reconfigure(struct fs_context *fc)
+{
+	struct nilfs_fs_context *ctx = fc->fs_private;
+	struct super_block *sb = fc->root->d_sb;
+	struct the_nilfs *nilfs = sb->s_fs_info;
+	int err;
+
+	sync_filesystem(sb);
+
+	err = -EINVAL;
+
+	if (!nilfs_valid_fs(nilfs)) {
+		nilfs_warn(sb,
+			   "couldn't remount because the filesystem is in an incomplete recovery state");
+		goto ignore_opts;
+	}
+	if ((bool)(fc->sb_flags & SB_RDONLY) == sb_rdonly(sb))
+		goto out;
+	if (fc->sb_flags & SB_RDONLY) {
+		sb->s_flags |= SB_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) {
+			nilfs_warn(sb,
+				   "couldn't remount RDWR because of unsupported optional features (%llx)",
+				   (unsigned long long)features);
+			err = -EROFS;
+			goto ignore_opts;
+		}
+
+		sb->s_flags &= ~SB_RDONLY;
+
+		root = NILFS_I(d_inode(sb->s_root))->i_root;
+		err = nilfs_attach_log_writer(sb, root);
+		if (err) {
+			sb->s_flags |= SB_RDONLY;
+			goto ignore_opts;
+		}
+
+		down_write(&nilfs->ns_sem);
+		nilfs_setup_super(sb, true);
+		up_write(&nilfs->ns_sem);
+	}
+ out:
+	sb->s_flags = (sb->s_flags & ~SB_POSIXACL);
+	/* Copy over parsed remount options */
+	nilfs->ns_mount_opt = ctx->ns_mount_opt;
+
+	return 0;
+
+ ignore_opts:
+	return err;
+}
+
+static int
+nilfs_get_tree(struct fs_context *fc)
+{
+	struct nilfs_fs_context *ctx = fc->fs_private;
+	struct super_block *s;
+	dev_t dev;
+	int err;
+
+	if (ctx->cno && !(fc->sb_flags & SB_RDONLY)) {
+		nilfs_err(NULL,
+			  "invalid option \"cp=%llu\": read-only option is not specified",
+			  ctx->cno);
+		return -EINVAL;
+	}
+
+	err = lookup_bdev(fc->source, &dev);
+	if (err)
+		return err;
+
+	s = sget_dev(fc, dev);
+	if (IS_ERR(s))
+		return PTR_ERR(s);
+
+	if (!s->s_root) {
+		err = setup_bdev_super(s, fc->sb_flags, fc);
+		if (!err)
+			err = nilfs_fill_super(s, fc);
+		if (err)
+			goto failed_super;
+
+		s->s_flags |= SB_ACTIVE;
+	} else if (!ctx->cno) {
+		if (nilfs_tree_is_busy(s->s_root)) {
+			if ((fc->sb_flags ^ s->s_flags) & SB_RDONLY) {
+				nilfs_err(s,
+					  "the device already has a %s mount.",
+					  sb_rdonly(s) ? "read-only" : "read/write");
+				err = -EBUSY;
+				goto failed_super;
+			}
+		} else {
+			/*
+			 * Try reconfigure to setup mount states if the current
+			 * tree is not mounted and only snapshots use this sb.
+			 *
+			 * Since nilfs_reconfigure() requires fc->root to be
+			 * set, set it first and release it on failure.
+			 */
+			fc->root = dget(s->s_root);
+			err = nilfs_reconfigure(fc);
+			if (err) {
+				dput(fc->root);
+				fc->root = NULL;  /* prevent double release */
+				goto failed_super;
+			}
+			return 0;
+		}
+	}
+
+	if (ctx->cno) {
+		struct dentry *root_dentry;
+
+		err = nilfs_attach_snapshot(s, ctx->cno, &root_dentry);
+		if (err)
+			goto failed_super;
+		fc->root = root_dentry;
+		return 0;
+	}
+
+	fc->root = dget(s->s_root);
+	return 0;
+
+ failed_super:
+	deactivate_locked_super(s);
+	return err;
+}
+
+static void nilfs_free_fc(struct fs_context *fc)
+{
+	kfree(fc->fs_private);
+}
+
+static const struct fs_context_operations nilfs_context_ops = {
+	.parse_param	= nilfs_parse_param,
+	.get_tree	= nilfs_get_tree,
+	.reconfigure	= nilfs_reconfigure,
+	.free		= nilfs_free_fc,
+};
+
+static int nilfs_init_fs_context(struct fs_context *fc)
+{
+	struct nilfs_fs_context *ctx;
+
+	ctx = kzalloc_obj(*ctx);
+	if (!ctx)
+		return -ENOMEM;
+
+	ctx->ns_mount_opt = NILFS_MOUNT_ERRORS_RO | NILFS_MOUNT_BARRIER;
+	fc->fs_private = ctx;
+	fc->ops = &nilfs_context_ops;
+
+	return 0;
+}
+
+struct file_system_type nilfs_fs_type = {
+	.owner    = THIS_MODULE,
+	.name     = "nilfs2",
+	.kill_sb  = kill_block_super,
+	.fs_flags = FS_REQUIRES_DEV,
+	.init_fs_context = nilfs_init_fs_context,
+	.parameters = nilfs_param_spec,
+};
+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
+	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();
+
+	kmem_cache_destroy(nilfs_inode_cachep);
+	kmem_cache_destroy(nilfs_transaction_cachep);
+	kmem_cache_destroy(nilfs_segbuf_cachep);
+	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|SLAB_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)