mms and rtsp patches = make modules fail

mms and rtsp patches = make modules fail

[Andy Furniss]

I don't really need them but I pressed Y :-) looks like some headers are 
missing from POM cvs of a couple of days old. mms and rtsp, unless my 
gcc 2.95.3 is being over pedantic. Vanilla 2.4.24 apart from 2IMQ IMQ 
nat and ESFQ patches.

In file included from ip_conntrack_standalone.c:28:
/home/andy/Sources/kernels/2.4.24/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack.h:74: 
field `ct_rtsp_info' has incomplete type
/home/andy/Sources/kernels/2.4.24/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack.h:89: 
field `ct_mms_info' has incomplete type
make[2]: *** [ip_conntrack_standalone.o] Error 1
make[2]: Leaving directory 
`/home/andy/Sources/kernels/2.4.24/linux-2.4.24/net/ipv4/netfilter'
make[1]: *** [_modsubdir_ipv4/netfilter] Error 2
make[1]: Leaving directory 
`/home/andy/Sources/kernels/2.4.24/linux-2.4.24/net

Andy.

Re: mms and rtsp patches = make modules fail

[Filip Sneppe]

On Wed, 2004-01-28 at 14:39, Andy Furniss wrote:
> I don't really need them but I pressed Y :-) looks like some headers are 
> missing from POM cvs of a couple of days old. mms and rtsp, unless my 
> gcc 2.95.3 is being over pedantic. Vanilla 2.4.24 apart from 2IMQ IMQ 
> nat and ESFQ patches.
> 
> In file included from ip_conntrack_standalone.c:28:
> /home/andy/Sources/kernels/2.4.24/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack.h:74: 
> field `ct_rtsp_info' has incomplete type
> /home/andy/Sources/kernels/2.4.24/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack.h:89: 
> field `ct_mms_info' has incomplete type
> make[2]: *** [ip_conntrack_standalone.o] Error 1
> make[2]: Leaving directory 
> `/home/andy/Sources/kernels/2.4.24/linux-2.4.24/net/ipv4/netfilter'
> make[1]: *** [_modsubdir_ipv4/netfilter] Error 2
> make[1]: Leaving directory 
> `/home/andy/Sources/kernels/2.4.24/linux-2.4.24/net
> 

Hi Andy,

I can't reproduce this with 2.4.24, some pending netfilter patches and
those two conntrackers. Where can I find those other two patches ?

Do other connection tracking/nat helpers from patch-o-matic compile
correctly, or did you just apply those two ?

If you want, you can send me a diff of your modified 2.4.24 against a
clean 2.4.24 and a description of what you've done/patched. I'll try
to see what goes wrong.

Regards,
Filip

Re: mms and rtsp patches = make modules fail

[Andy Furniss]

Filip Sneppe wrote:
> On Wed, 2004-01-28 at 14:39, Andy Furniss wrote:
> 
>>I don't really need them but I pressed Y :-) looks like some headers are 
>>missing from POM cvs of a couple of days old. mms and rtsp, unless my 
>>gcc 2.95.3 is being over pedantic. Vanilla 2.4.24 apart from 2IMQ IMQ 
>>nat and ESFQ patches.
>>
>>In file included from ip_conntrack_standalone.c:28:
>>/home/andy/Sources/kernels/2.4.24/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack.h:74: 
>>field `ct_rtsp_info' has incomplete type
>>/home/andy/Sources/kernels/2.4.24/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack.h:89: 
>>field `ct_mms_info' has incomplete type
>>make[2]: *** [ip_conntrack_standalone.o] Error 1
>>make[2]: Leaving directory 
>>`/home/andy/Sources/kernels/2.4.24/linux-2.4.24/net/ipv4/netfilter'
>>make[1]: *** [_modsubdir_ipv4/netfilter] Error 2
>>make[1]: Leaving directory 
>>`/home/andy/Sources/kernels/2.4.24/linux-2.4.24/net
>>
> 
> 
> Hi Andy,
> 
> I can't reproduce this with 2.4.24, some pending netfilter patches and
> those two conntrackers. Where can I find those other two patches ?
> 
> Do other connection tracking/nat helpers from patch-o-matic compile
> correctly, or did you just apply those two ?
> 
> If you want, you can send me a diff of your modified 2.4.24 against a
> clean 2.4.24 and a description of what you've done/patched. I'll try
> to see what goes wrong.
> 
> Regards,
> Filip
> 
> 
I just tested on a different machine, same old distro, though.

vanilla 2.4.24(I didn't patch this time) same POM (3 days old now) did 
./runme extra (last time I did make dep first)

and said y to connlimit nth quota u32 connbytes mms-conntrack-nat
rtsp-conntrack string and it said this -

Already applied: connlimit nth quota u32 connbytes conntrack-seqfile 
conntrack_arefcount conntrack_locking conntrack_nonat mms-conntrack-nat 
netfilter-docbook owner-socketlookup owner-supgids ownercmd 
rtsp-conntrack string

make mrproper (should I do this here?)
make menuconfig - said Y to filtering and m to everything in netfilter 
options
make dep
make bzImage
make modules

get the same error.

On the real compile I just backed out of the mms and rtsp did make 
modules again and it's running OK (though I haven't tested the new 
features yet).

Andy.

Re: mms and rtsp patches = make modules fail

[Filip Sneppe]

Hi,

On Thu, 2004-01-29 at 14:43, Andy Furniss wrote:
> I just tested on a different machine, same old distro, though.
> 
> vanilla 2.4.24(I didn't patch this time) same POM (3 days old now) did 
> ./runme extra (last time I did make dep first)
> 
> and said y to connlimit nth quota u32 connbytes mms-conntrack-nat
> rtsp-conntrack string and it said this -

With these patches, my 2.4.24 compiles fine when doing a 
"make SUBDIRS=net/ipv4/netfilter modules"
 
> Already applied: connlimit nth quota u32 connbytes conntrack-seqfile 
> conntrack_arefcount conntrack_locking conntrack_nonat mms-conntrack-nat 
> netfilter-docbook owner-socketlookup owner-supgids ownercmd 
> rtsp-conntrack string

Where do conntrack-seqfile, conntrack_arefcount, and these other
patches come from ? You must have selected them at one point in
time during a POM run, as they don't get applied automatically.

The list of patches mentioned in "Already applied: " step on
each other - in fact, in current CVS, POM warns against this and
one can only apply them all by forcing POM to apply some of them.

I get a compilation error afterwards, not the same as you, but
this probably depends on the order in which these patches have
been applied on our systems.

> make mrproper (should I do this here?)

It's ok to do a make mrproper here

Regards,
Filip

Re: mms and rtsp patches = make modules fail

[Andy Furniss]

Filip Sneppe wrote:
> Hi,
> 
> On Thu, 2004-01-29 at 14:43, Andy Furniss wrote:
> 
>>I just tested on a different machine, same old distro, though.
>>
>>vanilla 2.4.24(I didn't patch this time) same POM (3 days old now) did 
>>./runme extra (last time I did make dep first)
>>
>>and said y to connlimit nth quota u32 connbytes mms-conntrack-nat
>>rtsp-conntrack string and it said this -
> 
> 
> With these patches, my 2.4.24 compiles fine when doing a 
> "make SUBDIRS=net/ipv4/netfilter modules"
>  
> 
>>Already applied: connlimit nth quota u32 connbytes conntrack-seqfile 
>>conntrack_arefcount conntrack_locking conntrack_nonat mms-conntrack-nat 
>>netfilter-docbook owner-socketlookup owner-supgids ownercmd 
>>rtsp-conntrack string
> 
> 
> Where do conntrack-seqfile, conntrack_arefcount, and these other
> patches come from ? You must have selected them at one point in
> time during a POM run, as they don't get applied automatically.

I've just retested with a new cvs pulled iptables and patch-o-matic-ng.

They appear up to conntrack_nonat when I ./runme extra and select just 
connbytes, netfilter-docbook appears after mms-conntrack-nat and the 
others appear after saying n to the next option.
In fact if I do it again with a fresh kernel tree they appear even if I 
get no for everything.

I still get the error from fresh cvs and kernel tree, I do -

make mrproper
make dep
./runme extra in patch-o-matic-ng enter kernel dir (also tried export first)
just say y to connbytes mms rtsp.
make mrproper
make menuconfig y to network packet filtering
say m/y to everything except last two in netfilter options.
make dep
make SUBDIRS=net/ipv4/netfilter modules
error

Andy.

Re: mms and rtsp patches = make modules fail

[Filip Sneppe]

Hi Andy,

On Fri, 2004-01-30 at 13:44, Andy Furniss wrote:
> 
> I've just retested with a new cvs pulled iptables and patch-o-matic-ng.
> 
Ah, we may be on to something here. I have been using the old
patch-o-matic all this time.

Can you confirm that things work with the old pom runme script ? If so,
I'll look at what may be going wrong with patch-o-matic-ng.

Thanks,
Filip

Re: mms and rtsp patches = make modules fail

[Andy Furniss]

Filip Sneppe wrote:
> Hi Andy,
> 
> On Fri, 2004-01-30 at 13:44, Andy Furniss wrote:
> 
>>I've just retested with a new cvs pulled iptables and patch-o-matic-ng.
>>
> 
> Ah, we may be on to something here. I have been using the old
> patch-o-matic all this time.
> 
> Can you confirm that things work with the old pom runme script ? If so,
> I'll look at what may be going wrong with patch-o-matic-ng.
> 
> Thanks,
> Filip
> 
> 
> 

I tried with connbytes mms and rtsp - I got the chance to say n to the 
others this time, but got a different error.

ip_conntrack_rtsp.c:505: parse error before 
`this_object_must_be_defined_as_export_objs_in_the_Makefile' 
ip_conntrack_rtsp.c:505: warning: type defaults to `int' in declaration 
of `this_object_must_be_defined_as_export_objs_in_the_Makefile' 
ip_conntrack_rtsp.c:505: warning: data definition has no type or storage 
class make[1]: *** [ip_conntrack_rtsp.o] Error 1

Andy.