hppa and s390 build error, 1.2.9 + 2.4.24 headers

hppa and s390 build error, 1.2.9 + 2.4.24 headers

[Laurence J. Lane]

Does anyone have an idea of what is going wrong in the error below? It 
happens on s390 and hppa with 2.4.24 headers. Both build fine with 2.4.19
headers.

  full s390 build log: http://shorl.com/dufrustabulifra
  full hppa build log: http://shorl.com/fiprysabopesti
  all build logs: http://shorl.com/devufetiprahu

failed hppa build excerpt:

  cc -O2 -Wall -Wunused -I/build/buildd/iptables-1.2.9/debian/build/linux-2.4.24/include -Iinclude/ -DIPTABLES_VERSION=\"1.2.9\"  -fPIC -o extensions/libipt_connlimit_sh.o -c extensions/libipt_connlimit.c
  In file included from /build/buildd/iptables-1.2.9/debian/build/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack.h:52,
                   from extensions/libipt_connlimit.c:9:
  /build/buildd/iptables-1.2.9/debian/build/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack_icmp.h:9: error: parse error before "atomic_t"
  /build/buildd/iptables-1.2.9/debian/build/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack_icmp.h:9: warning: no semicolon at end of struct or union
  In file included from extensions/libipt_connlimit.c:9:
  /build/buildd/iptables-1.2.9/debian/build/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack.h:58: error: field `icmp' has incomplete type
  make[1]: *** [extensions/libipt_connlimit_sh.o] Error 1

Re: hppa and s390 build error, 1.2.9 + 2.4.24 headers

[Harald Welte]

[-- Attachment #1: Type: text/plain, Size: 1206 bytes --]

On Fri, Feb 13, 2004 at 12:42:22PM -0500, Laurence J. Lane wrote:
> Does anyone have an idea of what is going wrong in the error below? It 
> happens on s390 and hppa with 2.4.24 headers. Both build fine with 2.4.19
> headers.

Unfortunately I don't have access to any of those machines to further
debug the problem.

> failed hppa build excerpt:
> 
>   /build/buildd/iptables-1.2.9/debian/build/linux-2.4.24/include/linux/netfilter_ipv4/ip_conntrack_icmp.h:9: error: parse error before "atomic_t"

this sounds like some problem with atomic_t in an include file of a
userspace program (iptables in this case).

However, the typedef for atomic_t (asm-s390/atomic.h) looks quite fine
to me.

Is there anything special about kernel/userspace toolchain on those
architectures (like different compiler versions, ...)?

-- 
- Harald Welte <laforge@netfilter.org>             http://www.netfilter.org/
============================================================================
  "Fragmentation is like classful addressing -- an interesting early
   architectural error that shows how much experimentation was going
   on while IP was being designed."                    -- Paul Vixie

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: hppa and s390 build error, 1.2.9 + 2.4.24 headers

[Harald Welte]

[-- Attachment #1: Type: text/plain, Size: 955 bytes --]

On Tue, Feb 24, 2004 at 06:41:09PM -0500, Laurence J. Lane wrote:
 
> > Unfortunately I don't have access to any of those machines to further
> > debug the problem.
> 
> Debian affords many. Sign up to become a developer:

Sorry, but I'm definitely not going through the time-consuming process
of becoming a debian developer, just to hunt down a problem in the build
process of S/390.  

Either there is somebody seriously interested in getting this running on
S/390 and thus at least provide me access to such a machine, or the
issue is not that important after all.

-- 
- Harald Welte <laforge@netfilter.org>             http://www.netfilter.org/
============================================================================
  "Fragmentation is like classful addressing -- an interesting early
   architectural error that shows how much experimentation was going
   on while IP was being designed."                    -- Paul Vixie

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: hppa and s390 build error, 1.2.9 + 2.4.24 headers

[Laurence J. Lane]

On Sun, Feb 29, 2004 at 10:08:39PM +0100, Harald Welte wrote:

> > > Unfortunately I don't have access to any of those machines to
> > > further debug the problem.

> > Debian affords many. Sign up to become a developer:

> Sorry, but I'm definitely not going through the time-consuming
> process of becoming a debian developer, just to hunt down a
> problem in the build process of S/390.

I was mostly joking, but there are many benefits available to
developers and free software projects. It takes talented volunteers
to make it happen. You're still invited to sign up.

> Either there is somebody seriously interested in getting this
> running on S/390 and thus at least provide me access to such a
> machine, or the issue is not that important after all.

I copied the atomic_t typedef from asm/atomic.h to make it build. 

I suspect a gcc bug.