[Buildroot] ebtables and musl

[Peter Seiderer]

Hello Markus,

On Fri, 12 Jun 2020 13:10:28 -0700, Markus Mayer <mmayer at broadcom.com> wrote:

> Hi all,
>
> I see this patch in Buildroot:
> http://lists.busybox.net/pipermail/buildroot/2017-February/184225.html
>
> However, I am still running into the issue it is supposed to fix.
>
> This is with musl-1.2.0, the latest and greatest. I'm no longer stuck
> with the ancient libmusl that caused issues with procps. It doesn't
> seem to help with ebtables, though.
>
> /mnt/disk1/Development/br-musl/output/arm64/host/bin/aarch64-linux-gcc
> -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64  -Os
>   -fPIC -O3 -DPROGVERSION=\"2.0.10-4\" -DPROGNAME=\"ebtables\"
> -DPROGDATE=\"December\ 2011\" -D_PATH_ETHERTYPES=\"/etc/ethertypes\"
> -DEBTD_ARGC_MAX=50 -DEBTD_CMDLINE_MAXLN=2048
> -DLOCKFILE=\"/var/lib/ebtables/lock\" -DLOCKDIR=\"/var/lib/ebtables/\"
> -c -o extensions/ebt_802_3.o extensions/ebt_802_3.c -Iinclude/
> In file included from include/linux/netfilter_bridge.h:8:0,
>                  from include/linux/netfilter_bridge/ebtables.h:17,
>                  from include/ebtables_u.h:28,
>                  from useful_functions.c:25:
> include/linux/if_ether.h:119:8: error: redefinition of 'struct ethhdr'
>  struct ethhdr {
>         ^~~~~~
> In file included from
> ../../host/aarch64-buildroot-linux-musl/sysroot/usr/include/netinet/ether.h:8:0,
>                  from include/ebtables_u.h:27,
>                  from useful_functions.c:25:
> ../../host/aarch64-buildroot-linux-musl/sysroot/usr/include/netinet/if_ether.h:113:8:
> note: originally defined here
>  struct ethhdr {
>         ^~~~~~
> Makefile:78: recipe for target 'useful_functions.o' failed
> make[2]: *** [useful_functions.o] Error 1
> make[2]: *** Waiting for unfinished jobs....
>
> I am not quite sure what to make of it yet.
>
> Is it cause by the fact that include/linux/if_ether.h defines
>
> #ifndef _LINUX_IF_ETHER_H
> #define _LINUX_IF_ETHER_H
>
> but include/netinet/if_ether.h defines
>
> #ifndef _NETINET_IF_ETHER_H
> #define _NETINET_IF_ETHER_H
>
> and they both end up being included?
>
> $ find . -name if_ether.h -ls
>   6025818      4 -rw-r--r--   1 mmayer   mmayer       1506 May 26
> 16:30 ./build/linux-custom/include/linux/if_ether.h
>   6028683      8 -rw-r--r--   1 mmayer   mmayer       7461 May 26
> 16:30 ./build/linux-custom/include/uapi/linux/if_ether.h
>   6047054      8 -rw-r--r--   1 mmayer   mmayer       5751 Dec 15
> 2011 ./build/ebtables-2.0.10-4/include/linux/if_ether.h
>   5934557      8 -rw-r--r--   1 mmayer   mmayer       7016 Jun 10
> 16:52 ./host/aarch64-buildroot-linux-musl/sysroot/usr/include/linux/if_ether.h
>   5935035      4 -rw-r--r--   1 mmayer   mmayer       3625 Jun 10
> 16:52 ./host/aarch64-buildroot-linux-musl/sysroot/usr/include/netinet/if_ether.h
>
> Also, there's five different copies of if_ether.h floating around.

Care to provide a config/defconfig file for the failure (and provide the
used buildroot version)?

>
> While we are on the topic of ebtables, why is buildroot still using
> v2.0.10-4 from 2014 when there is 2.0.11 from 2019 (as per
> http://ftp.netfilter.org/pub/ebtables/)? Is it just a matter of
> submitting a patch to update it or is there an underlying reason to
> keep the old version around?

Probably not, care to send an ebtables update patch?

Regards,
Peter

>
> Thanks,
> -Markus
> _______________________________________________
> buildroot mailing list
> buildroot at busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot