Re: [RFC] ssl-bump security bugs

On 29/04/11 20:33, Tsantilas Christos wrote:
> On 04/29/2011 08:30 AM, Amos Jeffries wrote:
>> On 29/04/11 05:26, Tsantilas Christos wrote:
>>> On 04/12/2011 11:10 PM, Alex Rousskov wrote:
>>>> On 04/11/2011 11:33 PM, Amos Jeffries wrote:
>>>>> On 12/04/11 03:28, Alex Rousskov wrote:
>>>>>> On 04/10/2011 02:53 AM, Amos Jeffries wrote:
>>>>>>
>>> .....
>>>>>>> * The decrypted requests are not re-encrypted when sent outbound.
>>>>>>> IIRC
>>>>>>> there were measure attempted to make this happen, but they seem to
>>>>>>> have
>>>>>>> been unsuccessful.
>>>
>>> Do we have any such report? Which is the used configuration?
>>> I did some tests here, and also I tried to find such cases but I did not
>>> found. The traffic in my tests always re-encrypted before sent.
>>>
>>
>> I have two users mention replication of this in squid-users.
>>
>> The replicated case seems to be:
>> http_port ... ssl-bump
>> cache_peer ... parent ...
>> always_direct deny all
>
> Yep this is true.
> I think we should consider it as a bug.
> What I am actually getting is the following:
>
> <=https==> [proxy] <===http==>[proxy2]<==https==>
>

Yes that appears to be it. The problems being that a) proxy2 may not
have OpenSSL feature built in to re-crypt, and b) that the middle
channel is cleartext

>
> The easier solution is to add a check in FwdState::connectStart() in
> forward.cc file:
> if (fs->_peer && request->protocol == AnyP::PROTO_HTTPS) {
> anErr = errorCon();
> fail(enErr);
> return;
> }
> The above will just disable any ssl-bumped connection through any parent

As a short-term it may be okay. I don't see that working well with
normal proxied https:// URL requests, but those seem relatively rare.

>
> Also we should define if the "allow-direct" http-port options will have
> any effect in ssl-bump...

allow-direct being an accelerator mode option I think it will be
irrelevant once ssl-bump mode is created.

Since always_direct is documented as recommended, allow-direct behaviour
would seem to be implied for ssl-bump mode.

>
> The ideal I think is to use "CONNECT ..." when connecting through a
> parent proxy. But this is requires more development, I do not know if it
> is required and if we need it...

Yes. I imagine the MITM crowd doing interception of LAN traffic from
local workmates and funneling up to a corporate proxy will want that.

Amos

-- 
Please be using
   Current Stable Squid 2.7.STABLE9 or 3.1.12
   Beta testers wanted for 3.2.0.7 and 3.1.12.1