Re: exec and EXECSERVERS

[Thomas Bushnell, BSG]

prj@po.cwru.edu (Paul Jarc) writes:

> tb@becket.net (Thomas Bushnell, BSG) wrote:
> > We don't want to change other execs, because there is no reason to
> > think there is any kind of security implication for them.
> 
> Why not?  Doesn't ruid!=euid have the same implications as in Unix?
> (I.e., that a setuid program was executed, and no code since then has
> destroyed that information.)

Yes, but it's not a security issue any more, because no *augment* in
security is happening.

> In Unix, if I run setuid program foo, and foo runs program bar, then
> the dynamic loader, noticing that ruid!=euid, will ignore LD_PRELOAD,
> etc., when loading bar.  (Right?)  This is because LD_PRELOAD is under
> the control of a user different from the one whose privileges we have
> now.  Isn't the same true for the Hurd?

Well, we don't have LD_PRELOAD. :)