netdev - Re: [PATCH bpf-next 0/4] libbpf: add XDP binding support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <06177272-d6f1-c265-a0f8-91253c719a97@iogearbox.net>
Date:   Mon, 22 Jan 2018 15:00:50 +0100
From:   Daniel Borkmann <daniel@...earbox.net>
To:     Éric Leblond <eric@...it.org>
Cc:     Alexei Starovoitov <alexei.starovoitov@...il.com>,
        netdev@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH bpf-next 0/4] libbpf: add XDP binding support

On 01/20/2018 09:28 AM, Éric Leblond wrote:
> Hi,
> 
> Le 20 janv. 2018 9:21 AM, Daniel Borkmann <daniel@...earbox.net> a écrit :
> 
>     On 01/20/2018 03:27 AM, Alexei Starovoitov wrote:
>      > On Sat, Jan 20, 2018 at 03:00:37AM +0100, Daniel Borkmann wrote:
>      >> On 01/19/2018 12:43 AM, Eric Leblond wrote:
>      >>> Hello,
>      >>>
>      >>> This patchset rebases the libbpf code on latest bpf-next code and
>     addresses
>      >>> remarks by Daniel.
>      >>
>      >> Ok, I think it's a good start. We should later on clean up the
>      >> netlink handling code a bit, but that's all internal and can be
>      >> done in a second step. Applied to bpf-next, thanks Eric.
>      >
>      > Sorry, Eric, Daniel.
>      > I had to revert this patch set. It breaks build on systems
>      > where headers are not the most recent.
> 
> Oops, sorry.
> 
>      > Since libbpf is used by perf it has to be built cleanly on centos7 at least.
>      >
>      > The errors I got:
>      > bpf.c: In function ‘bpf_set_link_xdp_fd’:
>      > bpf.c:456:23: error: ‘SOL_NETLINK’ undeclared (first use in this function)
>      >   if (setsockopt(sock, SOL_NETLINK, NETLINK_EXT_ACK,
>      >                        ^~~~~~~~~~~
>      > bpf.c:456:23: note: each undeclared identifier is reported only once for
>     each function it appears in
>      > bpf.c:456:36: error: ‘NETLINK_EXT_ACK’ undeclared (first use in this
>     function)
>      >   if (setsockopt(sock, SOL_NETLINK, NETLINK_EXT_ACK,
>      >                                     ^~~~~~~~~~~~~~~
>      > nlattr.c: In function ‘nla_dump_errormsg’:
>      > nlattr.c:152:34: error: ‘NLMSGERR_ATTR_MAX’ undeclared (first use in this
>     function)
>      >   struct nla_policy extack_policy[NLMSGERR_ATTR_MAX + 1] = {
> 
>     Yeah, fully agree, thanks for catching this, Alexei!
> 
> What's the recommended solution here ? Include some kernel tree files or define 
> constant if not defined ?

I think typical way is to pull such headers into tools/include/.

Cheers,
Daniel