On Wed, 2008-12-17 at 20:53 +1100, James Morris wrote:
On Wed, 17 Dec 2008, Al Viro wrote:
> On Wed, Dec 17, 2008 at 06:24:40PM +1100, James Morris wrote:
> > On Wed, 17 Dec 2008, Al Viro wrote:
> >
> > > + struct {
> > > + uid_t uid;
> > > + gid_t gid;
> > > + mode_t mode;
> > > + u32 osid;
> > > + } ipc;
> >
> > 'osid' should be converted into 'secid' someday.
>
> Eh? Do you mean the field name there or the actual output? Either is
> trivial, of course, but the latter is up to userland folks and the
> former alone seems to be rather pointless...
I was thinking in terms of the kernel API, where 'secid' is the preferred
name for security identifiers ('sid' being an SELinux-specific term and
also conflicting with 'session id'). Given that it's exposed to userland,
I guess it's too late.
James meant just do s/osid/secid/ for continuity across the kernel (we
are trying to make the main kernel a bit more LSM agnostic and sid is an
SELinux term). The userspace exported part is actually a translated
string (I think we use ocontext= and scontext=).
There is no reason we couldn't do this in audit. But, I don't think
it's worth changing this patch, as I think audit refers to it as sid in
other places. Maybe I'll try to clean that up someday. I at least
added it to my "someday" todo list.
-Eric