netdev - Re: [PATCH v3 bpf-next 3/3] bpf: Add cb access in kfree

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>] [<thread-prev] [day] [month] [year] [list]

Date:   Thu, 7 Nov 2019 17:29:14 +0000
From:   Martin Lau <kafai@...com>
To:     Andrii Nakryiko <andrii.nakryiko@...il.com>
CC:     bpf <bpf@...r.kernel.org>, Networking <netdev@...r.kernel.org>,
        "Alexei Starovoitov" <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        "David Miller" <davem@...emloft.net>,
        Kernel Team <Kernel-team@...com>
Subject: Re: [PATCH v3 bpf-next 3/3] bpf: Add cb access in kfree_skb test

On Wed, Nov 06, 2019 at 06:50:02PM -0800, Andrii Nakryiko wrote:
> On Wed, Nov 6, 2019 at 5:47 PM Martin KaFai Lau <kafai@...com> wrote:
> >
> > Access the skb->cb[] in the kfree_skb test.
> >
> > Signed-off-by: Martin KaFai Lau <kafai@...com>
> > ---
> 
> Acked-by: Andrii Nakryiko <andriin@...com>
> 
> >  .../selftests/bpf/prog_tests/kfree_skb.c      | 54 +++++++++++++++----
> >  tools/testing/selftests/bpf/progs/kfree_skb.c | 25 +++++++--
> >  2 files changed, 63 insertions(+), 16 deletions(-)
> >
> 
> [...]
> 
> >
> > +       meta.ifindex = _(dev->ifindex);
I hit this when I put "meta.ifindex = dev->ifindex;" inside
__builtin_preserve_access_index:
libbpf: prog 'tp_btf/kfree_skb': relo #0: no matching targets found for [13] meta + 0:0
libbpf: prog 'tp_btf/kfree_skb': relo #0: failed to relocate: -3
libbpf: failed to perform CO-RE relocations: -3

My blind guess was it thought meta has to be relocated too.  Hence, I moved it
out such that I can put __builtin_preserve_access_index only to "dev->ifindex".

> > +       meta.cb8_0 = cb8[8];
> > +       meta.cb32_0 = cb32[2];
These two do not need relo.

> 
> Have you tried doing it inside __builtin_preserve_access_index region?
> Does it fail because of extra allocations against meta?
> 
> > +
> >         bpf_probe_read_kernel(&pkt_type, sizeof(pkt_type), _(&skb->__pkt_type_offset));
> >         pkt_type &= 7;
> >
> 
> [...]