* [Buildroot] [PATCH 1/1] libnet: explicitly pass libnet_cv_have_packet_socket=yes because PF_PACKET is always available on Linux
2015-11-18 12:15 [Buildroot] [PATCH 1/1] libnet: explicitly pass libnet_cv_have_packet_socket=yes because PF_PACKET is always available on Linux Joris Lijssens
@ 2015-11-19 22:00 ` Thomas Petazzoni
0 siblings, 0 replies; 2+ messages in thread
From: Thomas Petazzoni @ 2015-11-19 22:00 UTC (permalink / raw)
To: buildroot
Hello Joris,
Like your other patch, the commit title of this is too long. But
there's another issue, read below.
On Wed, 18 Nov 2015 13:15:30 +0100, Joris Lijssens wrote:
> Signed-off-by: Joris Lijssens <joris.lijssens@gmail.com>
> ---
> package/libnet/libnet.mk | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/package/libnet/libnet.mk b/package/libnet/libnet.mk
> index 3b62a79..7004417 100644
> --- a/package/libnet/libnet.mk
> +++ b/package/libnet/libnet.mk
> @@ -10,4 +10,9 @@ LIBNET_INSTALL_STAGING = YES
> LIBNET_LICENSE = BSD-2c, BSD-3c
> LIBNET_LICENSE_FILES = doc/COPYING
>
> +# PF_PACKET is always available on Linux, but build breaks for musl
> +ifneq ($(BR2_TOOLCHAIN_USES_MUSL),y)
> +LIBNET_CONF_OPTS = libnet_cv_have_packet_socket=yes
> +endif
In fact, even without defining libnet_cv_have_packet_socket=yes, libnet
does not build with musl. So either the build problem should be fixed
(it's a typical structure redefinition problem due to using the kernel
headers instead of the C library headers), but it is not super simple
to fix because libnet's code is somewhat convoluted.
I would suggest to simply mark the package as not available on musl
(and therefore netsniff-ng as well). Can you send a patch doing this,
and on top of this send a new version of this patch?
Thanks,
Thomas
--
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com
^ permalink raw reply [flat|nested] 2+ messages in thread