Re: pptp conntrack patch missing argument

Re: pptp conntrack patch missing argument

[Stephen Jones]

Hi Ali,

Thanks for the info!  I should be able to give this a spin, against a 
2.4.28 source tree, and test it in the next week or so!  Thanks again 
for pointing this out.  I didn't think of scanning the rest of the 
sources for calls to the same function in other sources, very good plan!

Best regards,

SJ

PS: I CC'd the netfilter list with your suggestion in the hopes that it 
does indeed fix the problem with the latest pom-ng release.

Ali wrote:

>Hi Stephen,
>
>While compiling a 2.4.27 kernel i suffered of the same problem you reported
>on the netfilter list and as you, searching on the web don't provide me any
>solution, but digging in the source i found that all calls to ip_nat_used_tuple
>in net/ipv4/netfilter/ip_nat_proto_* used the same convention:
>  if (!ip_nat_used_tuple(tuple, conntrack, maniptype, range->flags))
>changing line 108 in ip_nat_proto_gre.c to that make it compile cleanly. So far
>i haven't yet tested if it works ok but thougth that you may be interested :).
>
>Best regards,
>  Ali
>
>Pd: i'm not in the netfilter list (found your message in google) that's the reason to
>me writing you directly
>
>
>-------------------------------------------------
>Este mensaje fue enviado usando el servicio de correo en web de Infomed
>http://webmail.sld.cu
>
>
>  
>

Re: pptp conntrack patch missing argument

[Stephen Jones]

Hi Ali,

I was just in the process of sending you a note about this with the 
2.4.28 kernel.  I did a simple clean recompile with the pom-ng-20040621  
against iptables 1.2.11 source with only the pptp conntrack patches 
applied, and indeed, it did recompile without making the changes to the 
code in

line 108 in ip_nat_proto_gre.c.  I then later ran through and applied a boat-load more of the available patches, and ran back into the problem again!  I was about to start the tedious process of trying to figure out which patch nerfed pptp conntrack.  I definitely did apply both the tproxy and the nat-reservations patches, and after that, it was borked again!  I will cc the netfilter list with this info as well.  I was also unable to recompile iptables 1.2.11 after those patches, something borked ipt_mac.c in that series of patches.  I will pay closer attention next time, and see if I can narrow it down.  Thanks very much for the heads-up!


Ali wrote:

>Hi Stephen,
>Yesterday i proceed to compile another 2.4.27 kernel identical to the previous
>that give me the missing argument error but without the tproxy patch and it
>compiles without an error!!!! Curious about that i researched a little more
>in the code and found that the real culprit is nat-reservations, fortunately
>it seems that the only patch that depends on it is tproxy so if you no
>include it the probability of a clean compile is almost 100%
>  
>

>  
>
>>PS: I CC'd the netfilter list with your suggestion
>>    
>>
>thanks for that :)
>
>  
>
>>in the hopes that it does indeed fix the problem with the latest pom-ng
>>release.
>>    
>>
>I don't think that would be possible, from my research nat-reservations fix
>the protocols included in the standard kernel (icmp, tcp, udp) but have no
>way to fix other protocols not included (ex: gre), perhaps the only solution
>would be that the gre patch maintainer include a note about the problem and
>its solution.
>
>Best regards,
>  Ali
>
>
>
>-------------------------------------------------
>Este mensaje fue enviado usando el servicio de correo en web de Infomed
>http://webmail.sld.cu
>
>
>  
>