On 12/8/2013 4:13 AM, Dash Four wrote:
> Tom Eastep wrote:
>>> 4. Integrate "postcompile" and document it?
>>>     
>>
>> I assume that 'postcompile' would be invoked just before the compiler
>> generates the ip[6]tables input?
>>   
> No, it is currently invoked when the firewall file is created, but just 
> before it is passed for execution to ip[6]tables. I use this file to 
> apply 2 very ugly hacks to define my ifbX policy using ipset's ematch.

Okay -- I had forgotten that I had implemented that for you. I've now
documented it at http://www.shorewall.org/shorewall_extension_scripts.htm.

> 
>>> 5. Implement tc's "ematch" capability (so that the "ipset" ematch could 
>>> be used)?
>>>     
>>
>> I assume that this means that you would like to be able to define
>> 'basic' filters with ematch in the tcfilters file?
>>   
> Yes. If memory serves, ipset's ematch could be applied to any tc 
> statement and that includes most tc* files. Ipset's ematch is a powerful 
> tool and I can define complex rules (even more complex than a single 
> iptables statement), something like:
> 
> basic match (ipset"(set1 src)" or ipset"(set2 src)" or ipset"(set3 src)" 
> or ipset"(set4 src)") and ipset"(set5 dst)"

According to tc-ematch(8), ematch applies only to basic and flow filters.
> 
>>> 6. Implement IPSETs everywhere where iptables allows it, and I mean 
>>> *everywhere*? One example - I currently have matches inserted by 
>>> customised statements from my "started" file for some of the main chains 
>>> (like fw2zone and zone2fw), substituting the "net=xxx" option, but that 
>>> is one hell of an ugly hack and very prone to errors! The new-ish 
>>> version of the ipset match allow for byte and packet counters to be 
>>> used, so that could come handy in the accounting features in shorewall.
>>>     
>>
>> What is the barrier that prevents you from using INLINE ';' matches for
>> this?
>>   
> I can't. The "net=xxx" option for example applies to multiple chains 
> (fw2net, net2fw and so on), so I can't just slot INLINE statement in as 
> "net=xxx" has a wider scope.

Can you use INLINE in an action and then invoke the action in fw2net,
net2fw, etc?
> 
>>> 7. Implement access to the RAW tables/chains, similar to that of "rules"
>>
>> Which raw-table targets would you hope to be able to utilize? Could we
>> not just add suppport for them to the conntrack file?
>>   
> Don't know. What I currently have in my raw table is a sequence of 
> custom-defined chains (2 for logging, 5 for traversing, which are 
> interface-dependent) and in each of the traversing chains I have a 
> series of conditional DROP statements.

Okay -- so adding DROP and LOG support in the conntrack file would meet
your current needs?

Thanks,
-Tom
-- 
Tom Eastep        \ When I die, I want to go like my Grandfather who
Shoreline,         \ died peacefully in his sleep. Not screaming like
Washington, USA     \ all of the passengers in his car
http://shorewall.net \________________________________________________

Attachment: signature.asc
Description: OpenPGP digital signature

------------------------------------------------------------------------------
Sponsored by Intel(R) XDK 
Develop, test and display web and hybrid apps with a single code base.
Download it for free now!
http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
_______________________________________________
Shorewall-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/shorewall-devel

Reply via email to