Re: [PATCH net-next v4 3/4] bpf: BPF for lightweight tunnel infrastructure

[Daniel Borkmann <daniel@iogearbox.net>]

On 11/30/2016 05:10 PM, Thomas Graf wrote:
> Registers new BPF program types which correspond to the LWT hooks:
>    - BPF_PROG_TYPE_LWT_IN   => dst_input()
>    - BPF_PROG_TYPE_LWT_OUT  => dst_output()
>    - BPF_PROG_TYPE_LWT_XMIT => lwtunnel_xmit()
>
> The separate program types are required to differentiate between the
> capabilities each LWT hook allows:
>
>   * Programs attached to dst_input() or dst_output() are restricted and
>     may only read the data of an skb. This prevent modification and
>     possible invalidation of already validated packet headers on receive
>     and the construction of illegal headers while the IP headers are
>     still being assembled.
>
>   * Programs attached to lwtunnel_xmit() are allowed to modify packet
>     content as well as prepending an L2 header via a newly introduced
>     helper bpf_skb_change_head(). This is safe as lwtunnel_xmit() is
>     invoked after the IP header has been assembled completely.
[...]
>
> Signed-off-by: Thomas Graf <tgraf@suug.ch>

LGTMAFAICT, so:

Acked-by: Daniel Borkmann <daniel@iogearbox.net>

For the verifier change in may_access_direct_pkt_data(), would be
great if you could later on follow up with a selftest-suite case,
one where BPF_PROG_TYPE_LWT_IN/OUT prog tries to write and fails,
and one where BPF_PROG_TYPE_LWT_IN/OUT prog uses pkt data to pass
to helpers, for example, so that we can keep testing it when future
changes in that area are made. Thanks.