Re: Is starting a user program on priv port via inetd dangerous ?

[mrg () mame mu OZ AU (matthew green)]

[ ... ]
> A quick perusal of (4.3BSD) inetd shows that it forks, the child
> gets setuid & setgid to the user that ircd is supposed
> to run as (dougmc in this case), and exec()d.  Doesn't
> look too bad, but I just glanced at the code, and I couldn't
> say if any other version of UNIX doesn't do something dumb in inetd.

i don't think you understood what doug was asking.  i think running
ircd from inetd this was is "safe" in the context doug was talking
about - i don't think you can connect() out from a socket that is
passed to a process from inetd(), as inetd() has already done the
accept(), etc.

i'd have to write a test case to be sure, though.

> So, if there's a hole in ircd, it could cetainly be exploited as dougmc
> but probably not as root.  So it's probably not much worse than
> regular port 6667 in that respect.

it would be extremely hard for ircd to have a bug that would give
you anything but the chance of a denial of service attack.  it
just don't interact with unix (except to write a log file, perhaps),
besides using the tcp (and udp) layers.  last i checked, ircd exited
almost immediately if you ran it as root, or as suid root.

> It's still a pretty stupid idea, but you're already ware of that.

what, running an ircd, or running an ircd from inetd?

.mrg.