Messages in this thread | | | Date | Wed, 17 Sep 1997 17:33:12 +0100 (BST) | From | Chris Evans <> | Subject | Re: Process accounting question |
| |
On Wed, 17 Sep 1997, Theodore Y. Ts'o wrote:
> Date: Tue, 16 Sep 1997 22:55:37 +0100 (BST) > From: Chris Evans <chris@ferret.lmh.ox.ac.uk> > > BSD process accounting has aq flag which notes whether a process has used > super user privs or not. Should it count if a process uses super user > privs to override file access permissions? Currently we don't take this > into account. > > As long as everyone was careful to observe the convention that the > suser() test should be called at the *end* of logical OR or AND > statement, i.e.: > > if ((current->euid != uid) && !suser()) > return -EPERM; > > .... then it would be a very simple matter of changing the suser() > command to set the "superuser privs used flag". > > - Ted
That's what we do. But with suser() and NOT fsuser(). I was wondering if fsuser() should be treated similarly.
There are a couple of places where suser() ISN'T called last currently. I sent Linus a patch.
Chris
| |