> In http://acl.bestbits.at/man/extattr.5.html there is a statement:
>
> Device special files cannot be associated with extended user attributes
>
> What is the reason for this limitation? Why should there be a difference
> between regular files/directories and special files (device files)?
This limitation has been introduced since allowing extended user
attributes for special files leads to an awkward semantic mess: Extended
user attributes are in a sense treaded like file contents (concerning
permissions, for example). Now the "contents" of a special file are a
device, really. The device isn't even located on the filesystem the
special file is on. And the device doesn't have extended attributes.
It is possible to have devices on read-only file systems, which are
readable and writeable. It would not be possible to have extended
attributes in that case.
> I am also thinking in terms of HSM application (or DMAPI if you want). Where
> do you want HSM attributes to be placed? I thought it should be in trusted,
> because we might need access to them from user space. Other option is system
> (that would require accessing them from kernel code) or user (might be
> problematic, since regular user with write permission might remove them...
I am not sure about this. Just note that XFS and current ext2/ext3 don't
yet have the owner and trusted namespaces.
--Andreas.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/