dlmfs.c

ocfs1.4.1 oracle分布式文件系统

};

static struct inode *dlmfs_get_root_inode(struct super_block *sb)
{
	struct inode *inode = new_inode(sb);
	int mode = S_IFDIR | 0755;
	struct dlmfs_inode_private *ip;

	if (inode) {
		ip = DLMFS_I(inode);

		inode->i_mode = mode;
		inode->i_uid = current->fsuid;
		inode->i_gid = current->fsgid;
		inode->i_blocks = 0;
		inode->i_mapping->backing_dev_info = &dlmfs_backing_dev_info;
		inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
		inc_nlink(inode);

		inode->i_fop = &simple_dir_operations;
		inode->i_op = &dlmfs_root_inode_operations;
	}

	return inode;
}

static struct inode *dlmfs_get_inode(struct inode *parent,
				     struct dentry *dentry,
				     int mode)
{
	struct super_block *sb = parent->i_sb;
	struct inode * inode = new_inode(sb);
	struct dlmfs_inode_private *ip;

	if (!inode)
		return NULL;

	inode->i_mode = mode;
	inode->i_uid = current->fsuid;
	inode->i_gid = current->fsgid;
	inode->i_blocks = 0;
	inode->i_mapping->backing_dev_info = &dlmfs_backing_dev_info;
	inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;

	ip = DLMFS_I(inode);
	ip->ip_dlm = DLMFS_I(parent)->ip_dlm;

	switch (mode & S_IFMT) {
	default:
		/* for now we don't support anything other than
		 * directories and regular files. */
		BUG();
		break;
	case S_IFREG:
		inode->i_op = &dlmfs_file_inode_operations;
		inode->i_fop = &dlmfs_file_operations;

		i_size_write(inode,  DLM_LVB_LEN);

		user_dlm_lock_res_init(&ip->ip_lockres, dentry);

		/* released at clear_inode time, this insures that we
		 * get to drop the dlm reference on each lock *before*
		 * we call the unregister code for releasing parent
		 * directories. */
		ip->ip_parent = igrab(parent);
		BUG_ON(!ip->ip_parent);
		break;
	case S_IFDIR:
		inode->i_op = &dlmfs_dir_inode_operations;
		inode->i_fop = &simple_dir_operations;

		/* directory inodes start off with i_nlink ==
		 * 2 (for "." entry) */
		inc_nlink(inode);
		break;
	}

	if (parent->i_mode & S_ISGID) {
		inode->i_gid = parent->i_gid;
		if (S_ISDIR(mode))
			inode->i_mode |= S_ISGID;
	}

	return inode;
}

/*
 * File creation. Allocate an inode, and we're done..
 */
/* SMP-safe */
static int dlmfs_mkdir(struct inode * dir,
		       struct dentry * dentry,
		       int mode)
{
	int status;
	struct inode *inode = NULL;
	struct qstr *domain = &dentry->d_name;
	struct dlmfs_inode_private *ip;
	struct dlm_ctxt *dlm;
	struct dlm_protocol_version proto = user_locking_protocol;

	mlog(0, "mkdir %.*s\n", domain->len, domain->name);

	/* verify that we have a proper domain */
	if (domain->len >= O2NM_MAX_NAME_LEN) {
		status = -EINVAL;
		mlog(ML_ERROR, "invalid domain name for directory.\n");
		goto bail;
	}

	inode = dlmfs_get_inode(dir, dentry, mode | S_IFDIR);
	if (!inode) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}

	ip = DLMFS_I(inode);

	dlm = user_dlm_register_context(domain, &proto);
	if (IS_ERR(dlm)) {
		status = PTR_ERR(dlm);
		mlog(ML_ERROR, "Error %d could not register domain \"%.*s\"\n",
		     status, domain->len, domain->name);
		goto bail;
	}
	ip->ip_dlm = dlm;

	inc_nlink(dir);
	d_instantiate(dentry, inode);
	dget(dentry);	/* Extra count - pin the dentry in core */

	status = 0;
bail:
	if (status < 0)
		iput(inode);
	return status;
}

static int dlmfs_create(struct inode *dir,
			struct dentry *dentry,
			int mode,
			struct nameidata *nd)
{
	int status = 0;
	struct inode *inode;
	struct qstr *name = &dentry->d_name;

	mlog(0, "create %.*s\n", name->len, name->name);

	/* verify name is valid and doesn't contain any dlm reserved
	 * characters */
	if (name->len >= USER_DLM_LOCK_ID_MAX_LEN ||
	    name->name[0] == '$') {
		status = -EINVAL;
		mlog(ML_ERROR, "invalid lock name, %.*s\n", name->len,
		     name->name);
		goto bail;
	}

	inode = dlmfs_get_inode(dir, dentry, mode | S_IFREG);
	if (!inode) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}

	d_instantiate(dentry, inode);
	dget(dentry);	/* Extra count - pin the dentry in core */
bail:
	return status;
}

static int dlmfs_unlink(struct inode *dir,
			struct dentry *dentry)
{
	int status;
	struct inode *inode = dentry->d_inode;

	mlog(0, "unlink inode %lu\n", inode->i_ino);

	/* if there are no current holders, or none that are waiting
	 * to acquire a lock, this basically destroys our lockres. */
	status = user_dlm_destroy_lock(&DLMFS_I(inode)->ip_lockres);
	if (status < 0) {
		mlog(ML_ERROR, "unlink %.*s, error %d from destroy\n",
		     dentry->d_name.len, dentry->d_name.name, status);
		goto bail;
	}
	status = simple_unlink(dir, dentry);
bail:
	return status;
}

static int dlmfs_fill_super(struct super_block * sb,
			    void * data,
			    int silent)
{
	struct inode * inode;
	struct dentry * root;

	sb->s_maxbytes = MAX_LFS_FILESIZE;
	sb->s_blocksize = PAGE_CACHE_SIZE;
	sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
	sb->s_magic = DLMFS_MAGIC;
	sb->s_op = &dlmfs_ops;
	inode = dlmfs_get_root_inode(sb);
	if (!inode)
		return -ENOMEM;

	root = d_alloc_root(inode);
	if (!root) {
		iput(inode);
		return -ENOMEM;
	}
	sb->s_root = root;
	return 0;
}

#ifdef FOP_IS_NOT_CONST
static struct file_operations dlmfs_file_operations = {
#else
static const struct file_operations dlmfs_file_operations = {
#endif
	.open		= dlmfs_file_open,
	.release	= dlmfs_file_release,
	.read		= dlmfs_file_read,
	.write		= dlmfs_file_write,
};

#ifdef IOP_IS_NOT_CONST
static struct inode_operations dlmfs_dir_inode_operations = {
#else
static const struct inode_operations dlmfs_dir_inode_operations = {
#endif
	.create		= dlmfs_create,
	.lookup		= simple_lookup,
	.unlink		= dlmfs_unlink,
};

/* this way we can restrict mkdir to only the toplevel of the fs. */
#ifdef IOP_IS_NOT_CONST
static struct inode_operations dlmfs_root_inode_operations = {
#else
static const struct inode_operations dlmfs_root_inode_operations = {
#endif
	.lookup		= simple_lookup,
	.mkdir		= dlmfs_mkdir,
	.rmdir		= simple_rmdir,
};

#ifdef SOP_IS_NOT_CONST
static struct super_operations dlmfs_ops = {
#else
static const struct super_operations dlmfs_ops = {
#endif
	.statfs		= simple_statfs,
	.alloc_inode	= dlmfs_alloc_inode,
	.destroy_inode	= dlmfs_destroy_inode,
	.clear_inode	= dlmfs_clear_inode,
	.drop_inode	= generic_delete_inode,
};

#ifdef IOP_IS_NOT_CONST
static struct inode_operations dlmfs_file_inode_operations = {
#else
static const struct inode_operations dlmfs_file_inode_operations = {
#endif
	.getattr	= simple_getattr,
};

#ifdef NO_VFSMOUNT_IN_GET_SB_BDEV
static struct super_block *dlmfs_get_sb(struct file_system_type *fs_type,
					int flags, const char *dev_name,
					void *data)
{
	return get_sb_nodev(fs_type, flags, data, dlmfs_fill_super);
}
#else
static int dlmfs_get_sb(struct file_system_type *fs_type,
	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
{
	return get_sb_nodev(fs_type, flags, data, dlmfs_fill_super, mnt);
}
#endif

static struct file_system_type dlmfs_fs_type = {
	.owner		= THIS_MODULE,
	.name		= "ocfs2_dlmfs",
	.get_sb		= dlmfs_get_sb,
	.kill_sb	= kill_litter_super,
};

static int __init init_dlmfs_fs(void)
{
	int status;
	int cleanup_inode = 0, cleanup_worker = 0;

	dlmfs_print_version();

	dlmfs_inode_cache = kapi_kmem_cache_create("dlmfs_inode_cache",
				sizeof(struct dlmfs_inode_private),
				0, (SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|
					SLAB_MEM_SPREAD),
				dlmfs_init_once);
	if (!dlmfs_inode_cache)
		return -ENOMEM;
	cleanup_inode = 1;

	user_dlm_worker = create_singlethread_workqueue("user_dlm");
	if (!user_dlm_worker) {
		status = -ENOMEM;
		goto bail;
	}
	cleanup_worker = 1;

	status = register_filesystem(&dlmfs_fs_type);
bail:
	if (status) {
		if (cleanup_inode)
			kmem_cache_destroy(dlmfs_inode_cache);
		if (cleanup_worker)
			destroy_workqueue(user_dlm_worker);
	} else
		printk("OCFS2 User DLM kernel interface loaded\n");
	return status;
}

static void __exit exit_dlmfs_fs(void)
{
	unregister_filesystem(&dlmfs_fs_type);

	flush_workqueue(user_dlm_worker);
	destroy_workqueue(user_dlm_worker);

	kmem_cache_destroy(dlmfs_inode_cache);
}

MODULE_AUTHOR("Oracle");
MODULE_LICENSE("GPL");

module_init(init_dlmfs_fs)
module_exit(exit_dlmfs_fs)