build error when compiling nfsim against 2.6.23.17

build error when compiling nfsim against 2.6.23.17

[Fred Moyer]

I'm developing a connection tracking module for the 2.6.23.17 kernel
using the nf_conntrack helper examples (actually porting a nat module
I had for 2.4.30) and am trying to use the nfsim test framework
instead of my previous crude method of sending packets through the
machine.

I checked out a copy of nfsim and nfsim-testsuite from netfilter trunk
svn, and configured nfism with the command './configure
--kerneldir=../linux-2.6.23.17'.  However I encountered an error when
running make.  Should I be looking for an older tag of nfsim?  I'm
guessing that I have a version conflict which is causing this failure.
 I've also inlined my  gcc and make version (centos 5.2 linux
environment).

[phred@aircloud nfsim]$ make
doc/gen-help tools/config.c
Faking up config-help:config
gcc -Wmissing-prototypes -Wstrict-prototypes -Wunused -Wall -ggdb
-Wa,-W -I. -I/home/phred/SL-Kernel/nfsim/core
-I/home/phred/SL-Kernel/nfsim/kernelenv/include
-I/home/phred/SL-Kernel/nfsim/netfilter/include  -c -o tools/config.o
tools/config.c; ret=$?; rm -f `basename tools/config.c .c`-help*; exit
$ret
In file included from tools/config.c:26:
/home/phred/SL-Kernel/nfsim/netfilter/include/linux/netfilter_ipv4.h:78:
error: conflicting types for 'ip_route_me_harder'
/home/phred/SL-Kernel/nfsim/core/ipv4/ipv4.h:711: error: previous
declaration of 'ip_route_me_harder' was here
/home/phred/SL-Kernel/nfsim/netfilter/include/linux/netfilter_ipv4.h:80:
error: expected '=', ',', ';', 'asm' or '__attribute__' before
'nf_ip_checksum'
make: *** [tools/config.o] Error 1


[phred@aircloud nfsim]$ gcc -v
Using built-in specs.
Target: i386-redhat-linux
Configured with: ../configure --prefix=/usr --mandir=/usr/share/man
--infodir=/usr/share/info --enable-shared --enable-threads=posix
--enable-checking=release --with-system-zlib --enable-__cxa_atexit
--disable-libunwind-exceptions --enable-libgcj-multifile
--enable-languages=c,c++,objc,obj-c++,java,fortran,ada
--enable-java-awt=gtk --disable-dssi --enable-plugin
--with-java-home=/usr/lib/jvm/java-1.4.2-gcj-1.4.2.0/jre
--with-cpu=generic --host=i386-redhat-linux
Thread model: posix
gcc version 4.1.2 20071124 (Red Hat 4.1.2-42)
[phred@aircloud nfsim]$ make -v
GNU Make 3.81
Copyright (C) 2006  Free Software Foundation, Inc.
This is free software; see the source for copying conditions.
There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE.

This program built for i686-redhat-linux-gnu

Re: build error when compiling nfsim against 2.6.23.17

[Pablo Neira Ayuso]

Fred Moyer wrote:
> I'm developing a connection tracking module for the 2.6.23.17 kernel
> using the nf_conntrack helper examples (actually porting a nat module
> I had for 2.4.30) and am trying to use the nfsim test framework
> instead of my previous crude method of sending packets through the
> machine.

Sorry, nfsim currently lacks of maintainer.

-- 
"Los honestos son inadaptados sociales" -- Les Luthiers

Re: build error when compiling nfsim against 2.6.23.17

[Fred Moyer]

On Thu, Mar 5, 2009 at 4:25 AM, Pablo Neira Ayuso <pablo@netfilter.org> wrote:
> Fred Moyer wrote:
>> I'm developing a connection tracking module for the 2.6.23.17 kernel
>> using the nf_conntrack helper examples (actually porting a nat module
>> I had for 2.4.30) and am trying to use the nfsim test framework
>> instead of my previous crude method of sending packets through the
>> machine.
>
> Sorry, nfsim currently lacks of maintainer.

Ok, I'll make an effort to fix that issue and submit a patch.

What is the recommended alternative for testing connection helper modules?

Re: build error when compiling nfsim against 2.6.23.17

[Jan Engelhardt]

On Friday 2009-03-06 18:02, Fred Moyer wrote:
>On Thu, Mar 5, 2009 at 4:25 AM, Pablo Neira Ayuso <pablo@netfilter.org> wrote:
>> Fred Moyer wrote:
>>> I'm developing a connection tracking module for the 2.6.23.17 kernel
>>> using the nf_conntrack helper examples (actually porting a nat module
>>> I had for 2.4.30) and am trying to use the nfsim test framework
>>> instead of my previous crude method of sending packets through the
>>> machine.
>>
>> Sorry, nfsim currently lacks of maintainer.
>
>Ok, I'll make an effort to fix that issue and submit a patch.
>
>What is the recommended alternative for testing connection helper modules?

Testing them in a real world - perhaps virtual machines though - setup :-)

Re: build error when compiling nfsim against 2.6.23.17

[Fred Moyer]

On Fri, Mar 6, 2009 at 10:31 AM, Jan Engelhardt <jengelh@medozas.de> wrote:
>
> On Friday 2009-03-06 18:02, Fred Moyer wrote:
>>On Thu, Mar 5, 2009 at 4:25 AM, Pablo Neira Ayuso <pablo@netfilter.org> wrote:
>>> Fred Moyer wrote:
>>>> I'm developing a connection tracking module for the 2.6.23.17 kernel
>>>> using the nf_conntrack helper examples (actually porting a nat module
>>>> I had for 2.4.30) and am trying to use the nfsim test framework
>>>> instead of my previous crude method of sending packets through the
>>>> machine.
>>>
>>> Sorry, nfsim currently lacks of maintainer.
>>
>>Ok, I'll make an effort to fix that issue and submit a patch.
>>
>>What is the recommended alternative for testing connection helper modules?
>
> Testing them in a real world - perhaps virtual machines though - setup :-)

I've had great success developing and testing with VMWare Fusion on a
mac laptop.  The reboot time on an oops is maybe two minutes, about
the same as my metal testing server, except lacking the fan noise and
ability to live on my laptop.

I looked through the nfsim unit tests - very nice design, reminds me
of a lot of the perl code testing I do with Test::More.  It seems like
a testing approach that would lead to faster development than pushing
packets.  Normally to test I 'tail -f /var/log/kernel', and have
tcpdump running, but if I oops then I lose all that state.

Re: build error when compiling nfsim against 2.6.23.17

[Jan Engelhardt]

On Friday 2009-03-06 19:37, Fred Moyer wrote:
>>>
>>>What is the recommended alternative for testing connection helper modules?
>>
>> Testing them in a real world - perhaps virtual machines though - setup :-)
>
>I've had great success developing and testing with VMWare Fusion on a
>mac laptop.  The reboot time on an oops is maybe two minutes, about
>the same as my metal testing server, except lacking the fan noise and
>ability to live on my laptop.

At least VMware Workstation has the Snapshot feature which cuts the
"reboot" time down to whatever it takes to reread the disk image from disk.

>I looked through the nfsim unit tests - very nice design, reminds me
>of a lot of the perl code testing I do with Test::More.  It seems like
>a testing approach that would lead to faster development than pushing
>packets.  Normally to test I 'tail -f /var/log/kernel', and have
>tcpdump running, but if I oops then I lose all that state.

Avoiding oopses is the best strategy, at which point you can reedit-
compile-reload the module as often as you need without reverting to
a snapshot.
>