On Mon, 29 Jul 2019 18:46:25 -0700
Michael Forney <[email protected]> wrote:

> On 2019-07-27, Mattias Andrée <[email protected]> wrote:
> > A file is executable only if the effective user
> > have permission to execute it. The real user's
> > permissions do not matter.  
> 
> Thanks for the patch, but doesn't this only make a difference if the
> `which` binary itself is setuid? If not, can you provide an example
> that is fixed by this patch?
> 
> I looked at a few other implementations and they just use access(3),
> which behaves like faccessat(3) with no flags.
> 

setuid is inherited (exec(3) never changes the effective user according
to POSIX unless the new process have the setuid flag and it is not blocked
by the ST_NOSUID mount option). However, I cannot think of a real world
scenario where this would matter; it would be if the user have a program
similar to sudo that only changes the effective user.

Reply via email to