> > 1) you really _should_ be doing such explicit access checks _anyway_. > > > For kernel objects? If the access your check grants is the same as > what the kernel grants, you're wasting your time, and if it's not, then > you've introduced race conditions. for object == file/dir, you're absol. right.