Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

[Emmanuel Guiton]

Hi!

In the target module I'm writing, I need to customize the values 
ip_ct_tcp_timeout_/somthing/ in the ip_conntrack_proto_tcp.c file. I 
tried to write an extern function on the model of ip_ct_refresh, but I 
always got the same problem with depmod:

depmod: *** Unresolved symbols in 
/lib/modules/2.4.23/kernel/net/ipv4/netfilter/ipt_MY_TARGET.o
depmod:     ip_ct_proto_tcp_set_timeouts

Does someone have a good advice on how to proceed correctly?

                 Emmanuel

Re: Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

[Henrik Nordstrom]

On Wed, 14 Jan 2004, Emmanuel Guiton wrote:

> In the target module I'm writing, I need to customize the values 
> ip_ct_tcp_timeout_/somthing/ in the ip_conntrack_proto_tcp.c file. I 
> tried to write an extern function on the model of ip_ct_refresh, but I 
> always got the same problem with depmod:
> 
> depmod: *** Unresolved symbols in 
> /lib/modules/2.4.23/kernel/net/ipv4/netfilter/ipt_MY_TARGET.o
> depmod:     ip_ct_proto_tcp_set_timeouts
> 
> Does someone have a good advice on how to proceed correctly?

Looks like you have not exported this function from ip_conntrack. To use a 
function from another module than it is defined in the function must first 
be exported from the module where it is defined.

Regards
Henrik

Re: Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

[Emmanuel Guiton]

Henrik Nordstrom wrote:

>On Wed, 14 Jan 2004, Emmanuel Guiton wrote:
>
>  
>
>>In the target module I'm writing, I need to customize the values 
>>ip_ct_tcp_timeout_/somthing/ in the ip_conntrack_proto_tcp.c file. I 
>>tried to write an extern function on the model of ip_ct_refresh, but I 
>>always got the same problem with depmod:
>>
>>depmod: *** Unresolved symbols in 
>>/lib/modules/2.4.23/kernel/net/ipv4/netfilter/ipt_MY_TARGET.o
>>depmod:     ip_ct_proto_tcp_set_timeouts
>>
>>Does someone have a good advice on how to proceed correctly?
>>    
>>
>
>Looks like you have not exported this function from ip_conntrack. To use a 
>function from another module than it is defined in the function must first 
>be exported from the module where it is defined.
>
>Regards
>Henrik
>  
>

Well, I think I did. I defined my function in the 
ip_conntrack_proto_tcp.c and I exported its prototype in ip_conntrack.h.
Then I call it in my target module which also uses ip_conntrack.h.
Should I have done otherwise?

        Emmanuel

Re: Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

[KOVACS Krisztian]

  Hi,

On Wed, 2004-01-14 at 13:39, Emmanuel Guiton wrote:
> >Looks like you have not exported this function from ip_conntrack. To use a 
> >function from another module than it is defined in the function must first 
> >be exported from the module where it is defined.
> 
> Well, I think I did. I defined my function in the 
> ip_conntrack_proto_tcp.c and I exported its prototype in ip_conntrack.h.
> Then I call it in my target module which also uses ip_conntrack.h.

  You have to explicitly export it using the EXPORT_SYMBOL() (or
EXPORT_SYMBOL_GPL) macro in ip_conntrack_standalone.c. It has a lot of
other exported symbols, just add another one.

-- 
 Regards,
   Krisztian KOVACS

Re: Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

[Emmanuel Guiton]

KOVACS Krisztian wrote:

>  Hi,
>
>On Wed, 2004-01-14 at 13:39, Emmanuel Guiton wrote:
>  
>
>>>Looks like you have not exported this function from ip_conntrack. To use a 
>>>function from another module than it is defined in the function must first 
>>>be exported from the module where it is defined.
>>>      
>>>
>>Well, I think I did. I defined my function in the 
>>ip_conntrack_proto_tcp.c and I exported its prototype in ip_conntrack.h.
>>Then I call it in my target module which also uses ip_conntrack.h.
>>    
>>
>
>  You have to explicitly export it using the EXPORT_SYMBOL() (or
>EXPORT_SYMBOL_GPL) macro in ip_conntrack_standalone.c. It has a lot of
>other exported symbols, just add another one.
>
>  
>
Right, it's working now.
Thanks!

           Emmanuel

Side effect? - Re: Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

[Emmanuel Guiton]

Hi!

I was a bit optimistic in my last e-mail. Compilation is totally ok, 
insertion of the modules in the kernel seems to go fine... until I want 
to use them. The nat table does not work anymore.
The ip_tables module seems to be there:
pc104:# iptables -L   
Chain INPUT (policy ACCEPT)
target     prot opt source               destination        

Chain FORWARD (policy ACCEPT)
target     prot opt source               destination        

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination  

But if I try the nat table:
pc104:# iptables -t nat -L
modprobe: Can't locate module ip_tables
iptables v1.2.9: can't initialize iptables table `nat': Table does not 
exist (do you need to insmod?)
Perhaps iptables or your kernel needs to be upgraded.

If I do not add my function in ip_conntrack_standalone.c, everything's 
ok (well except that I can't use my funciton of course).
To be very precise I just added the following:

In ip_conntrack_standalone.c:
EXPORT_SYMBOL(ip_ct_proto_tcp_set_timeouts);

In ip_conntrack.h:
extern void ip_ct_proto_tcp_set_timeous(/datatypes/);

In ip_conntrack_proto_tcp.c:
void ip_ct_proto_tcp_set_tcp_timeouts(unsigned long new_value, ...)
{
    ...
    ip_ct_tcp_timeout_established = new_value SECS;
    ...
}

And the calls to the function in ipt_MY_TARGET.c:
ip_ct_proto_tcp_set_timeouts(/values/);


What do I do wrong?

           Emmanuel



Emmanuel Guiton wrote:

> KOVACS Krisztian wrote:
>
>>  Hi,
>>
>> On Wed, 2004-01-14 at 13:39, Emmanuel Guiton wrote:
>>  
>>
>>>> Looks like you have not exported this function from ip_conntrack. 
>>>> To use a function from another module than it is defined in the 
>>>> function must first be exported from the module where it is defined.
>>>>     
>>>
>>> Well, I think I did. I defined my function in the 
>>> ip_conntrack_proto_tcp.c and I exported its prototype in 
>>> ip_conntrack.h.
>>> Then I call it in my target module which also uses ip_conntrack.h.
>>>   
>>
>>
>>  You have to explicitly export it using the EXPORT_SYMBOL() (or
>> EXPORT_SYMBOL_GPL) macro in ip_conntrack_standalone.c. It has a lot of
>> other exported symbols, just add another one.
>>
>>  
>>
> Right, it's working now.
> Thanks!
>
>           Emmanuel
>

Re: Side effect? - Re: Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

[KOVACS Krisztian]

  Hi,

On Thu, 2004-01-15 at 09:26, Emmanuel Guiton wrote:
> I was a bit optimistic in my last e-mail. Compilation is totally ok, 
> insertion of the modules in the kernel seems to go fine... until I want 
> to use them. The nat table does not work anymore.
> The ip_tables module seems to be there:
> pc104:# iptables -L   
> Chain INPUT (policy ACCEPT)
> target     prot opt source               destination        
> 
> Chain FORWARD (policy ACCEPT)
> target     prot opt source               destination        
> 
> Chain OUTPUT (policy ACCEPT)
> target     prot opt source               destination  
> 
> But if I try the nat table:
> pc104:# iptables -t nat -L
> modprobe: Can't locate module ip_tables
> iptables v1.2.9: can't initialize iptables table `nat': Table does not 
> exist (do you need to insmod?)
> Perhaps iptables or your kernel needs to be upgraded.

  Are you sure iptable_nat is loaded in the kernel? And what does your
kernel log say?

-- 
 Regards,
   Krisztian KOVACS

Userspace msg, hw_addr unused?

[Scott MacKay]

I was looking thru the userspace code for material
passed up in the ipq_packet_msg structure, looking for
any data from the ethernet header.  I noticed that the
hw_addr array never seems to be filled out, even
though hw_addrlen does.  Any reason for that?

Also, if I wanted to patch the kernel's ip_queue.c to
pass up the source and destination ether address, any
ideas on where I get the packet's data from?  Thanks!!

__________________________________
Do you Yahoo!?
Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes
http://hotjobs.sweepstakes.yahoo.com/signingbonus

Re: Userspace msg, hw_addr unused?

[Scott MacKay]

In poking around, I think I see what would be needed. 
Any confirmation would be great.
entry->skb has a mac.ethernet structure.  This looks
to be a ethhdr, so it has a h_source and h_dest.
2 options seem to be to add a ethdr to the
ipq_packet_msg and memcpy the entry->skb->mac.ethernet
or to add 2 fields (maybe removing the hw_addr if
unused) and memcpy the h_source and h_dest.

I think, if I alter those fields, though, that there
is alot of recompile needed.  Is there a better way to
get to the ethernet fields?


--- Scott MacKay <scottmackay@yahoo.com> wrote:
> I was looking thru the userspace code for material
> passed up in the ipq_packet_msg structure, looking
> for
> any data from the ethernet header.  I noticed that
> the
> hw_addr array never seems to be filled out, even
> though hw_addrlen does.  Any reason for that?
> 
> Also, if I wanted to patch the kernel's ip_queue.c
> to
> pass up the source and destination ether address,
> any
> ideas on where I get the packet's data from? 
> Thanks!!
> 
> __________________________________
> Do you Yahoo!?
> Yahoo! Hotjobs: Enter the "Signing Bonus"
> Sweepstakes
> http://hotjobs.sweepstakes.yahoo.com/signingbonus
> 


__________________________________
Do you Yahoo!?
Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes
http://hotjobs.sweepstakes.yahoo.com/signingbonus

Re: Side effect? - Re: Changing the content of tcp timeouts in ip_conntrack_proto_tcp.c

[Emmanuel Guiton]

KOVACS Krisztian wrote:

>  Are you sure iptable_nat is loaded in the kernel? And what does your
>kernel log say?
>
>  
>
Well, as far as I've modified my sources I can't reproduce the error 
immediately. But a good guess would be that the module was not loaded 
and more specifically the problem may simply occured in the ip_conntrack 
module, thus avoiding iptable_nat to work.
But the problem is now solved. I have not changed anything except the 
name of my function which is much shorter than my original one. Now it's 
working, even the execution produces the result I was expecting. Isn't 
there a specific bound on the function's name? The first name I used was 
44 characters long (it was in fact a longer version of the one I 
indicated in my previous mail).

Tanks for your help anyway.

          Emmanuel

Last minute erratum: I'm stupid. - Re: Side effect? - Re: Changing the content of tcp timeouts ...

[Emmanuel Guiton]

Emmanuel Guiton wrote:

> KOVACS Krisztian wrote:
>
>>  Are you sure iptable_nat is loaded in the kernel? And what does your
>> kernel log say?
>>
>>  
>>
> Well, as far as I've modified my sources I can't reproduce the error 
> immediately. But a good guess would be that the module was not loaded 
> and more specifically the problem may simply occured in the 
> ip_conntrack module, thus avoiding iptable_nat to work.
> But the problem is now solved. I have not changed anything except the 
> name of my function which is much shorter than my original one. Now 
> it's working, even the execution produces the result I was expecting. 
> Isn't there a specific bound on the function's name? The first name I 
> used was 44 characters long (it was in fact a longer version of the 
> one I indicated in my previous mail).
>
> Tanks for your help anyway.
>
>          Emmanuel
>
I just noticed the real cause of the problem. A syntax error in the name 
of my function. I thought I had checked it... Please, forget about all 
the previous messages.
Mea culpa.

           Emmanuel