[lsb-discuss] statfs specification addition: review

Camp, TracyX E tracyx.e.camp at intel.com
Thu Dec 21 14:52:25 PST 2006

>Well, the xattr interface is currently shared between JFS, ext3, XFS,
>and REiserfs.  There are different restrictions about the size of the
>supported extended attribute names and values, so depending on what
>you are doing you may still need to be aware of which filesystem you
>are using, but at least the core API is the same, and if you're only
>storing 8 or 16 bytes worth of xattr, you're not going to have a
>serious problem.  (If you're going to store several megabytes worth of
>rootkit for example --- which seems to the primary use of NTFS streams
>on Windows at the moment --- then you will need to be aware of what
>filesystem you are using.)

No, it just argues that xattr max length is something that the system
should be able to tell me about via pathconf() or similar at run-time.

>I will note, BTW, that the ext2 filesystem format actually has been
>around longer than NTFS has ever been in existence.  New features have
>been added, sure, but the original filesystem format has still been
>supported, and I don't see that changing anytime soon.  And if you
>claim that ext2 isn't being used because it's been replaced by newer
>filesystems with more functionality, well, the same can be said about
>FAT; and FAT isn't even backwards compatible with NTFS.

This is entirely orthoginal and mostly full of red-herring, but an
interesting historical debate none the less.  ext2 and NTFS both came
about in 1993 (January and July respectively), but yes ext2 has been
largely replaced in production with file systems that report a different
magic (reiserfs for example).  I think my point still stands, that the
file system is emphimeral over a long-enough time span and applications
should not be asked to depend upon it specifically.

>Anyway, before we go into the effort to standardize xattr and ACL
>interfaces --- do people care?  Are there applications who would be
>interested in coding against the LSB if we added those interfaces?
>And if those aren't the filesystem interests, what are?  I've heard
>Direct I/O and locking semantics --- what else?  And what are the
>locking issues that people are most worried about?

As I said, I think this is something that should be systematically
approached by considering the VFS interface in detail and not
hoge-podged together as people ask for things.

Tracy Camp

>					- Ted

More information about the lsb-discuss mailing list