namei.c

ocfs1.4.1 oracle分布式文件系统

		status = ocfs2_create_new_inode_locks(inode);
		if (status < 0)
			mlog_errno(status);
	}

	status = 0; /* error in ocfs2_create_new_inode_locks is not
		     * critical */

	*ret_inode = inode;
leave:
	if (status < 0) {
		if (*new_fe_bh) {
			brelse(*new_fe_bh);
			*new_fe_bh = NULL;
		}
		if (inode)
			iput(inode);
	}

	mlog_exit(status);
	return status;
}

static int ocfs2_mkdir(struct inode *dir,
		       struct dentry *dentry,
		       int mode)
{
	int ret;

	mlog_entry("(0x%p, 0x%p, %d, '%.*s')\n", dir, dentry, mode,
		   dentry->d_name.len, dentry->d_name.name);
	ret = ocfs2_mknod(dir, dentry, mode | S_IFDIR, 0);
	mlog_exit(ret);

	return ret;
}

static int ocfs2_create(struct inode *dir,
			struct dentry *dentry,
			int mode,
			struct nameidata *nd)
{
	int ret;

	mlog_entry("(0x%p, 0x%p, %d, '%.*s')\n", dir, dentry, mode,
		   dentry->d_name.len, dentry->d_name.name);
	ret = ocfs2_mknod(dir, dentry, mode | S_IFREG, 0);
	mlog_exit(ret);

	return ret;
}

static int ocfs2_link(struct dentry *old_dentry,
		      struct inode *dir,
		      struct dentry *dentry)
{
	handle_t *handle;
	struct inode *inode = old_dentry->d_inode;
	int err;
	struct buffer_head *fe_bh = NULL;
	struct buffer_head *parent_fe_bh = NULL;
	struct buffer_head *de_bh = NULL;
	struct ocfs2_dinode *fe = NULL;
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);

	mlog_entry("(inode=%lu, old='%.*s' new='%.*s')\n", inode->i_ino,
		   old_dentry->d_name.len, old_dentry->d_name.name,
		   dentry->d_name.len, dentry->d_name.name);

	if (S_ISDIR(inode->i_mode))
		return -EPERM;

	err = ocfs2_inode_lock(dir, &parent_fe_bh, 1);
	if (err < 0) {
		if (err != -ENOENT)
			mlog_errno(err);
		return err;
	}

	if (!dir->i_nlink) {
		err = -ENOENT;
		goto out;
	}

	err = ocfs2_check_dir_for_entry(dir, dentry->d_name.name,
					dentry->d_name.len);
	if (err)
		goto out;

	err = ocfs2_prepare_dir_for_insert(osb, dir, parent_fe_bh,
					   dentry->d_name.name,
					   dentry->d_name.len, &de_bh);
	if (err < 0) {
		mlog_errno(err);
		goto out;
	}

	err = ocfs2_inode_lock(inode, &fe_bh, 1);
	if (err < 0) {
		if (err != -ENOENT)
			mlog_errno(err);
		goto out;
	}

	fe = (struct ocfs2_dinode *) fe_bh->b_data;
	if (le16_to_cpu(fe->i_links_count) >= OCFS2_LINK_MAX) {
		err = -EMLINK;
		goto out_unlock_inode;
	}

	handle = ocfs2_start_trans(osb, OCFS2_LINK_CREDITS);
	if (IS_ERR(handle)) {
		err = PTR_ERR(handle);
		handle = NULL;
		mlog_errno(err);
		goto out_unlock_inode;
	}

	err = ocfs2_journal_access(handle, inode, fe_bh,
				   OCFS2_JOURNAL_ACCESS_WRITE);
	if (err < 0) {
		mlog_errno(err);
		goto out_commit;
	}

	inc_nlink(inode);
	inode->i_ctime = CURRENT_TIME;
	fe->i_links_count = cpu_to_le16(inode->i_nlink);
	fe->i_ctime = cpu_to_le64(inode->i_ctime.tv_sec);
	fe->i_ctime_nsec = cpu_to_le32(inode->i_ctime.tv_nsec);

	err = ocfs2_journal_dirty(handle, fe_bh);
	if (err < 0) {
		le16_add_cpu(&fe->i_links_count, -1);
		drop_nlink(inode);
		mlog_errno(err);
		goto out_commit;
	}

	err = ocfs2_add_entry(handle, dentry, inode,
			      OCFS2_I(inode)->ip_blkno,
			      parent_fe_bh, de_bh);
	if (err) {
		le16_add_cpu(&fe->i_links_count, -1);
		drop_nlink(inode);
		mlog_errno(err);
		goto out_commit;
	}

	err = ocfs2_dentry_attach_lock(dentry, inode, OCFS2_I(dir)->ip_blkno);
	if (err) {
		mlog_errno(err);
		goto out_commit;
	}

	atomic_inc(&inode->i_count);
	dentry->d_op = &ocfs2_dentry_ops;
	d_instantiate(dentry, inode);

out_commit:
	ocfs2_commit_trans(osb, handle);
out_unlock_inode:
	ocfs2_inode_unlock(inode, 1);

out:
	ocfs2_inode_unlock(dir, 1);

	if (de_bh)
		brelse(de_bh);
	if (fe_bh)
		brelse(fe_bh);
	if (parent_fe_bh)
		brelse(parent_fe_bh);

	mlog_exit(err);

	return err;
}

/*
 * Takes and drops an exclusive lock on the given dentry. This will
 * force other nodes to drop it.
 */
static int ocfs2_remote_dentry_delete(struct dentry *dentry)
{
	int ret;

	ret = ocfs2_dentry_lock(dentry, 1);
	if (ret)
		mlog_errno(ret);
	else
		ocfs2_dentry_unlock(dentry, 1);

	return ret;
}

static inline int inode_is_unlinkable(struct inode *inode)
{
	if (S_ISDIR(inode->i_mode)) {
		if (inode->i_nlink == 2)
			return 1;
		return 0;
	}

	if (inode->i_nlink == 1)
		return 1;
	return 0;
}

static int ocfs2_unlink(struct inode *dir,
			struct dentry *dentry)
{
	int status;
	int child_locked = 0;
	struct inode *inode = dentry->d_inode;
	struct inode *orphan_dir = NULL;
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
	u64 blkno;
	struct ocfs2_dinode *fe = NULL;
	struct buffer_head *fe_bh = NULL;
	struct buffer_head *parent_node_bh = NULL;
	handle_t *handle = NULL;
	struct ocfs2_dir_entry *dirent = NULL;
	struct buffer_head *dirent_bh = NULL;
	char orphan_name[OCFS2_ORPHAN_NAMELEN + 1];
	struct buffer_head *orphan_entry_bh = NULL;

	mlog_entry("(0x%p, 0x%p, '%.*s')\n", dir, dentry,
		   dentry->d_name.len, dentry->d_name.name);

	BUG_ON(dentry->d_parent->d_inode != dir);

	mlog(0, "ino = %llu\n", (unsigned long long)OCFS2_I(inode)->ip_blkno);

	if (inode == osb->root_inode) {
		mlog(0, "Cannot delete the root directory\n");
		return -EPERM;
	}

	status = ocfs2_inode_lock(dir, &parent_node_bh, 1);
	if (status < 0) {
		if (status != -ENOENT)
			mlog_errno(status);
		return status;
	}

	status = ocfs2_find_files_on_disk(dentry->d_name.name,
					  dentry->d_name.len, &blkno,
					  dir, &dirent_bh, &dirent);
	if (status < 0) {
		if (status != -ENOENT)
			mlog_errno(status);
		goto leave;
	}

	if (OCFS2_I(inode)->ip_blkno != blkno) {
		status = -ENOENT;

		mlog(0, "ip_blkno %llu != dirent blkno %llu ip_flags = %x\n",
		     (unsigned long long)OCFS2_I(inode)->ip_blkno,
		     (unsigned long long)blkno, OCFS2_I(inode)->ip_flags);
		goto leave;
	}

	status = ocfs2_inode_lock(inode, &fe_bh, 1);
	if (status < 0) {
		if (status != -ENOENT)
			mlog_errno(status);
		goto leave;
	}
	child_locked = 1;

	if (S_ISDIR(inode->i_mode)) {
	       	if (!ocfs2_empty_dir(inode)) {
			status = -ENOTEMPTY;
			goto leave;
		} else if (inode->i_nlink != 2) {
			status = -ENOTEMPTY;
			goto leave;
		}
	}

	status = ocfs2_remote_dentry_delete(dentry);
	if (status < 0) {
		/* This remote delete should succeed under all normal
		 * circumstances. */
		mlog_errno(status);
		goto leave;
	}

	if (inode_is_unlinkable(inode)) {
		status = ocfs2_prepare_orphan_dir(osb, &orphan_dir, inode,
						  orphan_name,
						  &orphan_entry_bh);
		if (status < 0) {
			mlog_errno(status);
			goto leave;
		}
	}

	handle = ocfs2_start_trans(osb, OCFS2_UNLINK_CREDITS);
	if (IS_ERR(handle)) {
		status = PTR_ERR(handle);
		handle = NULL;
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_journal_access(handle, inode, fe_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	fe = (struct ocfs2_dinode *) fe_bh->b_data;

	if (inode_is_unlinkable(inode)) {
		status = ocfs2_orphan_add(osb, handle, inode, fe, orphan_name,
					  orphan_entry_bh, orphan_dir);
		if (status < 0) {
			mlog_errno(status);
			goto leave;
		}
	}

	/* delete the name from the parent dir */
	status = ocfs2_delete_entry(handle, dir, dirent, dirent_bh);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	if (S_ISDIR(inode->i_mode))
		drop_nlink(inode);
	drop_nlink(inode);
	fe->i_links_count = cpu_to_le16(inode->i_nlink);

	status = ocfs2_journal_dirty(handle, fe_bh);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	dir->i_ctime = dir->i_mtime = CURRENT_TIME;
	if (S_ISDIR(inode->i_mode))
		drop_nlink(dir);

	status = ocfs2_mark_inode_dirty(handle, dir, parent_node_bh);
	if (status < 0) {
		mlog_errno(status);
		if (S_ISDIR(inode->i_mode))
			inc_nlink(dir);
	}

leave:
	if (handle)
		ocfs2_commit_trans(osb, handle);

	if (child_locked)
		ocfs2_inode_unlock(inode, 1);

	ocfs2_inode_unlock(dir, 1);

	if (orphan_dir) {
		/* This was locked for us in ocfs2_prepare_orphan_dir() */
		ocfs2_inode_unlock(orphan_dir, 1);
		mutex_unlock(&orphan_dir->i_mutex);
		iput(orphan_dir);
	}

	if (fe_bh)
		brelse(fe_bh);

	if (dirent_bh)
		brelse(dirent_bh);

	if (parent_node_bh)
		brelse(parent_node_bh);

	if (orphan_entry_bh)
		brelse(orphan_entry_bh);

	mlog_exit(status);

	return status;
}

/*
 * The only place this should be used is rename!
 * if they have the same id, then the 1st one is the only one locked.
 */
static int ocfs2_double_lock(struct ocfs2_super *osb,
			     struct buffer_head **bh1,
			     struct inode *inode1,
			     struct buffer_head **bh2,
			     struct inode *inode2)
{
	int status;
	struct ocfs2_inode_info *oi1 = OCFS2_I(inode1);
	struct ocfs2_inode_info *oi2 = OCFS2_I(inode2);
	struct buffer_head **tmpbh;
	struct inode *tmpinode;

	mlog_entry("(inode1 = %llu, inode2 = %llu)\n",
		   (unsigned long long)oi1->ip_blkno,
		   (unsigned long long)oi2->ip_blkno);

	if (*bh1)
		*bh1 = NULL;
	if (*bh2)
		*bh2 = NULL;

	/* we always want to lock the one with the lower lockid first. */
	if (oi1->ip_blkno != oi2->ip_blkno) {
		if (oi1->ip_blkno < oi2->ip_blkno) {
			/* switch id1 and id2 around */
			mlog(0, "switching them around...\n");
			tmpbh = bh2;
			bh2 = bh1;
			bh1 = tmpbh;

			tmpinode = inode2;
			inode2 = inode1;
			inode1 = tmpinode;
		}
		/* lock id2 */
		status = ocfs2_inode_lock(inode2, bh2, 1);
		if (status < 0) {
			if (status != -ENOENT)
				mlog_errno(status);
			goto bail;
		}
	}

	/* lock id1 */
	status = ocfs2_inode_lock(inode1, bh1, 1);
	if (status < 0) {
		/*
		 * An error return must mean that no cluster locks
		 * were held on function exit.
		 */
		if (oi1->ip_blkno != oi2->ip_blkno)
			ocfs2_inode_unlock(inode2, 1);

		if (status != -ENOENT)
			mlog_errno(status);
	}

bail:
	mlog_exit(status);
	return status;
}

static void ocfs2_double_unlock(struct inode *inode1, struct inode *inode2)
{
	ocfs2_inode_unlock(inode1, 1);

	if (inode1 != inode2)
		ocfs2_inode_unlock(inode2, 1);
}

static int ocfs2_rename(struct inode *old_dir,
			struct dentry *old_dentry,
			struct inode *new_dir,
			struct dentry *new_dentry)
{
	int status = 0, rename_lock = 0, parents_locked = 0;
	int old_child_locked = 0, new_child_locked = 0;
	struct inode *old_inode = old_dentry->d_inode;
	struct inode *new_inode = new_dentry->d_inode;
	struct inode *orphan_dir = NULL;
	struct ocfs2_dinode *newfe = NULL;
	char orphan_name[OCFS2_ORPHAN_NAMELEN + 1];
	struct buffer_head *orphan_entry_bh = NULL;
	struct buffer_head *newfe_bh = NULL;
	struct buffer_head *old_inode_bh = NULL;
	struct buffer_head *insert_entry_bh = NULL;
	struct ocfs2_super *osb = NULL;
	u64 newfe_blkno, old_de_ino;
	handle_t *handle = NULL;
	struct buffer_head *old_dir_bh = NULL;