Hm, sounds like a bug in CIFS - it should return 0 if there are no xattrs.

I looked superficially in the code, the ENODATA seems to come from here https://elixir.bootlin.com/linux/latest/source/fs/cifs/smb2maperror.c#L380
So the error means "the remote fs has xattrs disabled" not that this particuliar file has no xattr. So this is approximately ENOTSUP.

Here they explicitly try to map -ENODATA for listxattr() to returning 0: https://elixir.bootlin.com/linux/latest/source/fs/cifs/smb2ops.c#L794

	/*
	 * If ea_name is NULL (listxattr) and there are no EAs, return 0 as it's
	 * not an error. Otherwise, the specified ea_name was not found.
	 */
	if (!rc)
		rc = move_smb2_ea_to_cifs(ea_data, buf_size, smb2_data,
					  SMB2_MAX_EA_BUF, ea_name);
	else if (!ea_name && rc == -ENODATA)
		rc = 0;

But something is clearly going wrong.

Ah, it's only very recently fixed - only in v4.18: torvalds/linux@ae2cd7f. Mystery solved then. Let's still keep the -ENODATA check so that older kernels still work, it doesn't cost us anything.