On Fri, 23 Feb 2001, Andreas Gruenbacher wrote:
> The other namespaces have other restictions:
> user = everybody who has access to the file data
> inode = owners only (I'm thinking of renaming this to owner,
> as Robert suggested.)
> system (or acl_posix etc.) = defined in the kernel
> trusted = defined by trusted user-space processes
What with a storm of mail, I haven't had a chance to respond to other
messages yet, but will sometime in the next day or two. However, I did
have a couple of comments on namespacing:
1) I'd like it if we moved to namespace partitioning not inside the name
string. I know that Andreas and I both used namespacing based on the
first character of the name string in some versions of our
implementations, but I actually think that is a bad idea now. I
believe you change this in your new interface, and I think that
regardless of other outcomes of the discussion of the new interface,
this is something to retain.
2) I'd actually like to propose an additional namespace, the "internal" or
"filesystem" namespace, to be used by the file system to manage
internal attributes, as distinguished from "system" attributes. For
example, suppose I implement ACLs by backing them into EAs. The
name for the EAs to store the flattened and fs/hardware/whateverelse-
specific data would be in the filsystem namespace. My feeling was
that "system" EAs would have a consistent exportable syntax/semantic
that the consumer would have to be aware of, but that potentially
it would be consistent regardless of file system. The filesystem
namespace would really be intended for internal use only, but would
allow the file system to reserve names for internal use without
any worries about namespace collisions. These names might not even
be accessible above the VFS layer, but the namespace identifier would
be reserved.
Robert N M Watson FreeBSD Core Team, TrustedBSD Project
[EMAIL PROTECTED] NAI Labs, Safeport Network Services
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]