DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 16F843858C41
Message-ID: <91ad2c64-153b-1692-e100-8adcba0bf828@aussiebb.com.au>
Date: Tue, 8 Aug 2023 11:40:55 +1000
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.13.0
Subject: Re: sshd_config AllowStreamLocalForwarding *remote not possible* /
 effectively privsep off
To: cygwin@cygwin.com
References: <883e0ae2-1ac7-1474-ba06-10d9de441390@aussiebb.com.au>
 <ZNEslrRdAQxRtVBD@calimero.vinschen.de>
 <db4eb850-bb65-dc66-eeb4-00ff49cb6777@aussiebb.com.au>
In-Reply-To: <db4eb850-bb65-dc66-eeb4-00ff49cb6777@aussiebb.com.au>
From: Shaddy Baddah via Cygwin <cygwin@cygwin.com>
Reply-To: Shaddy Baddah <lithium-cygwin@shaddybaddah.name>
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset="us-ascii"; Format="flowed"
Sender: "Cygwin" <cygwin-bounces+archive-cygwin=delorie.com@cygwin.com>

Hi,

I've just updated the subject line for accuracy. Only remote/reverse
unix socket forwarding fails.

Further, I have a clarification that might have significance:

On 8/08/2023 3:40 am, Corinna Vinschen via Cygwin wrote:
 > On Aug  7 22:11, Shaddy Baddah via Cygwin wrote:
..
 >
 >> DISABLE_FD_PASS is always set by autoconf for Cygwin. And my reading is
 >> that not having that capability effectively means whatever the other
 >> criteria, the executing process doesn't have sufficient "separation" of
 >> privilege to be treated in the same manner.

Perhaps contrary to expectation, with the more conventional
remote/reverse TCP port forwarding, with Cygwin sshd, the LISTEN port
exists in the, is it called the monitor
(http://www.citi.umich.edu/u/provos/ssh/priv.jpg)/intermediatary sshd
process.

So something like:

|>~C
|ssh> -R 12345:22

will result in a (confirmed by netstat) LISTEN port in the SYSTEM owned
sshd process, which is the parent of the non-privileged owned sshd
process.

I'm not suggesting that this is not a considered situation, because to
my knowledge, it's a much different situation allowing an ssh user to
manipulate the filesystem (for unix sockets), as SYSTEM. Than using
netsocks as SYSTEM to try and bind TCP ports... I think???

But it certainly aligns with my newfound understanding of Cygwin's
"trade-off" form of privilege separation.

-- 
Regards,
Shaddy


-- 
Problem reports:      https://cygwin.com/problems.html
FAQ:                  https://cygwin.com/faq/
Documentation:        https://cygwin.com/docs.html
Unsubscribe info:     https://cygwin.com/ml/#unsubscribe-simple