For use in embedded environments, it would be useful if IPv6 support could be
disabled via ./configure. This option was available in older versions, but then
removed.

I get an error when trying to attach a file:
URL: http://bugzilla.netfilter.org/attachment.cgi
undef error - Undefined subroutine Fh::slice at
data/template/template/en/default/global/hidden-fields.html.tmpl line 58 

So my proposed patch comes here, sorry the for inconvenience:


diff -ru iptables-1.4.2/configure.ac iptables-1.4.2_patched/configure.ac
--- iptables-1.4.2/configure.ac 2009-03-29 15:49:24.764299587 +0000
+++ iptables-1.4.2_patched/configure.ac 2009-03-29 15:38:29.082512186 +0000
@@ -37,6 +37,9 @@
    [enable_devel="$enableval"], [enable_devel="yes"])
 AC_ARG_ENABLE([libipq],
    AS_HELP_STRING([--enable-libipq], [Build and install libipq]))
+AC_ARG_ENABLE([ipv6],
+    AS_HELP_STRING([--enable-ipv6], [Build and install ip6tables]),
+    [enable_ipv6="$enableval"], [enable_ipv6="yes"])
 AC_ARG_WITH([pkgconfigdir], AS_HELP_STRING([--with-pkgconfigdir=PATH],
    [Path to the pkgconfig directory [[LIBDIR/pkgconfig]]]),
    [pkgconfigdir="$withval"], [pkgconfigdir='${libdir}/pkgconfig'])
@@ -53,6 +56,7 @@
 AM_CONDITIONAL([ENABLE_SHARED], [test "$enable_shared" = "yes"])
 AM_CONDITIONAL([ENABLE_DEVEL], [test "$enable_devel" = "yes"])
 AM_CONDITIONAL([ENABLE_LIBIPQ], [test "$enable_libipq" = "yes"])
+AM_CONDITIONAL([ENABLE_IPV6], [test "$enable_ipv6" == "yes"])

 regular_CFLAGS="-D_LARGEFILE_SOURCE=1 -D_LARGE_FILES -D_FILE_OFFSET_BITS=64 \
    -D_REENTRANT -Wall -Waggregate-return -Wmissing-declarations \
Only in iptables-1.4.2: configure.ac.orig
Only in iptables-1.4.2: configure.ac.rej
diff -ru iptables-1.4.2/Makefile.am iptables-1.4.2_patched/Makefile.am
--- iptables-1.4.2/Makefile.am  2008-10-08 16:31:58.000000000 +0000
+++ iptables-1.4.2_patched/Makefile.am  2009-03-29 15:38:29.082512186 +0000
@@ -80,11 +80,17 @@
 endif

 if ENABLE_STATIC
-sbin_PROGRAMS += iptables-static ip6tables-static
+sbin_PROGRAMS += iptables-static
+if ENABLE_IPV6
+sbin_PROGRAMS += ip6tables-static
+endif
+
 endif
 if ENABLE_SHARED
-sbin_PROGRAMS += iptables iptables-multi iptables-restore iptables-save \
-                 ip6tables ip6tables-multi ip6tables-restore ip6tables-save
+sbin_PROGRAMS += iptables iptables-multi iptables-restore iptables-save
+if ENABLE_IPV6
+sbin_PROGRAMS += ip6tables ip6tables-multi ip6tables-restore ip6tables-save
+endif
 endif

 iptables.8: ${srcdir}/iptables.8.in extensions/matches4.man
extensions/targets4.man
diff -ru iptables-1.4.2/xtables.c iptables-1.4.2_patched/xtables.c
--- iptables-1.4.2/xtables.c    2008-10-08 16:31:58.000000000 +0000
+++ iptables-1.4.2_patched/xtables.c    2009-03-29 15:38:29.082512186 +0000
@@ -964,6 +964,8 @@
    }
 }

+#ifdef ENABLE_IPV6
+
 const char *ip6addr_to_numeric(const struct in6_addr *addrp)
 {
    /* 0000:0000:0000:0000:0000:000.000.000.000
@@ -1189,6 +1191,7 @@
            }
    }
 }
+#endif

 void save_string(const char *value)
 {

Slightly incomplete (nowhere do you add -DENABLE_IPV6 or similar), but I will
fix that up.

Would not it be simpler to just provide the zero ipv6 address variable
outselves?

Ignore comment #3, it was sort of for bug #569.

What I do wonder, you only use #ifdef ENABLE_IPV6 in xtables.c, but there are
more places where it would have to be done, i.e. extensions/libxt_*.c. I am not
keen on sprinkling the code with lots of these ifdefs, though.
Any comments from Patrick?

For now, it has been committed as 8e58613df53f5f83e8ab92dec61d8065c68d967d and
a094eb0f2a57592b6f3cf42fdbb9d49fead2d57c, though these commits merely skip
building pure-v4/v6 parts; libxtables remains as is. Also see bug #569 which is
relevant to people wanting to disable components for embedded reasons.