UID/GID bug in chrooted shells fixed.

Tom Worley raq at worley.co.uk
Mon Jun 17 13:54:01 EST 2002

On Monday 17 June 2002 9:39 pm, you wrote:
> Uh, I suppose you could do it that assuming you don't mind the fact that
> there's bound to be zillions of ways for somebody with a real login on your
> machine to use it give themselves root access

Yup, of course...
However, the only users on the system will be root, and those stuck in the 
chroot jail on a noexec partition with only bash, su and rsync binaries that 
can be executed (using mount --bind -o ro). So, the only place that can be 
written to cannot execute files, unless you can see a way around this?

> but even with that rsync
> still might be confused.  If you'd rather modify rsync than write a
> wrapper, something that will be likely to cause rsync less confusion is
>     if (geteuid() == 0)
> 	setuid(0);
It appears to work fine the way I did it, but I'll try this way, thanks.
And yes, I'm a memeber of the mailing list, ya don't have to CC it to me ;-)

Tom Worley
Worley Web Solutions

More information about the rsync mailing list