netdev - [PATCH bpf v2 0/3] bpf: Rework bpf_redirect_neigh() to allow supplying nexthop from caller

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-next>] [day] [month] [year] [list]

Message-ID: <160319106111.15822.18417665895694986295.stgit@toke.dk>
Date:   Tue, 20 Oct 2020 12:51:01 +0200
From:   Toke Høiland-Jørgensen <toke@...hat.com>
To:     Daniel Borkmann <daniel@...earbox.net>
Cc:     David Ahern <dsahern@...nel.org>, netdev@...r.kernel.org,
        bpf@...r.kernel.org
Subject: [PATCH bpf v2 0/3] bpf: Rework bpf_redirect_neigh() to allow
 supplying nexthop from caller

Based on previous discussion[0], we determined that it would be beneficial to
rework bpf_redirect_neigh() so the caller can supply the nexthop information
(e.g., from a previous call to bpf_fib_lookup()). This way, the two helpers can
be combined without incurring a second FIB lookup to find the nexthop, and
bpf_fib_lookup() becomes usable even if no nexthop entry currently exists.

This patch (and accompanying selftest update) accomplishes this by way of an
optional paramter to bpf_redirect_neigh(). This series is against the -bpf tree,
since we need to change this call signature before it becomes API. As a
companion change, it also adds a flag to bpf_fib_lookup() that will make it skip
the neighbour lookup, for cases where the caller knows it is likely for fail
anyway, and wants to go straight to bpf_redirect_neigh().

[0] https://lore.kernel.org/bpf/393e17fc-d187-3a8d-2f0d-a627c7c63fca@iogearbox.net/

Changelog:

v2:
- Add 'unused' member to fill hole in bpf_redir_neigh struct (David Ahern)
- Fix compilation with INET/INET6 disabled - properly this time (kbot)
- Add back the BPF_FIB_LOOKUP_SKIP_NEIGH flag as new patch 2 (Daniel)

v1:
- Rebase on -bpf tree
- Fix compilation with INET/INET6 disabled (kbot)
- Keep v4/v6 signatures similar, use internal flag (Daniel)
- Use a separate selftest BPF program instead of modifying existing one (Daniel)
- Fix a few style nits (David Ahern)

---

Toke Høiland-Jørgensen (3):
      bpf_redirect_neigh: Support supplying the nexthop as a helper parameter
      bpf_fib_lookup: optionally skip neighbour lookup
      selftests: Update test_tc_redirect.sh to use the modified bpf_redirect_neigh()

 include/uapi/linux/bpf.h                      |  10 +-
 net/core/filter.c                             |  16 +-
 tools/include/uapi/linux/bpf.h                |  10 +-
 .../selftests/bpf/progs/test_tc_neigh.c       |   5 +-
 .../selftests/bpf/progs/test_tc_neigh_fib.c   | 153 ++++++++++++++++++
 .../testing/selftests/bpf/test_tc_redirect.sh |  18 ++-
 6 files changed, 197 insertions(+), 15 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/progs/test_tc_neigh_fib.c