iptables limit module causes an alignment error on a 64-bit kernel and 32-bit userland. After inserting a rule containing limit, any other operation results in the following message on kernel dmesg: x_tables: ip_tables: limit.0 match: invalid size 40 (kernel) != (user) 48 looking at the limit code, its the only one module that uses CONFIG_COMPAT for legacy code.
1) This isn't critical - if you want something fixed then you should at least be polite and don't think about your little corner only. Critical would be something minimal not building, for example a basic root filesystem with just busybox with a default configuration. Iptables is just one package among many, and the failure is a subset of the functionality of iptables on a subset of many architectures. 2) This is an upstream bug, unless there's a patch or knob to signal the iptables build about mixed 64/32 kernel/userland (which was present several releases ago IIRC) or fixing this issue then there's nothing we can do on our side alone - you should contact the netfilter devels about this issue since it's a bug.
I`m not Thinking about my "Little corner". Anyway I`m not submitting any patches anymore. Thanks for your "Attention"
On a second look i was somewhat rude, so sorry about that. This is a volunteer effort and people labeling problems as critical when they aren't is sometimes very draining, it's like people going all "me me me" and wanting to be the center of the universe or something - when in fact it'll probably get them ignored on a perpetual basis when they do that too much. Taking that into consideration most of the contributors don't have mips64 boards to test things on and the qemu emulation is mostly broken so only basic things can be tested (bare shell, can't do much), and only mips64 N32 ABI uses mixed 64/32-bit kernel/userspace from all the combinations supported at the moment, so we can't even reproduce this issue. Again, sorry, and i suggest you take it up with the netfilter dev team at their bugzilla or their mailing list (probably better). And please do provide them with details including kernel & iptables version.
I got it. Wasn't my intention to get attention when I marked it as a critical bug. Sorry about that. I really not cared to check as this failure is upstream, and not a buildroot fault. Thanks