On Tue, Mar 05, 2002 at 12:10:14PM -0500, Chandrasekhar KRISHNAN wrote:
> Why does Squid consider ports greater than 1024 safe, when it is easy to
> abuse them (since a non-root user can bind on them)?
>
> I went through the discussions about the safe ports in the mailing list,
> but couldn't find answer to this question. Any info will be appreciated.
It depends on your definition of "safe". Most Squid administrators will be
concerned about users using Squid to perform actions they would not normally
have permission to do, by connecting to machines that have trust
relationships with the Squid server. Usually security-critical applications
run on ports < 1024, precisely because these are the ports that cannot be
hijacked by non-root users. So "safe" in this context means ports that can
be connected to without resulting in unintended privilege elivation.
Of course, even port 80 can be unsafe if there happens to be an intranet web
server that trusts the Squid server, but such things cannot be solved by the
default config.
Personally, I always configure my Squid servers to use a blacklist of unsafe
ports, since that causes less user complaints about the vast armies of
webservers running on non-standard port numbers, but then our servers aren't
entangled in any complex trust relationships.
Adam
-- Adam Rice \ Web Technician \ Central Team \ Newsquest Digital Media XML + Unix + CGI + E-mail + DNS + HTTP + Security @ Phone 01254 691171Received on Tue Mar 05 2002 - 10:40:39 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:06:43 MST