netdev - pull-request: bpf 2018-05-18

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: <20180518002617.15231-1-daniel@iogearbox.net>
Date:   Fri, 18 May 2018 02:26:17 +0200
From:   Daniel Borkmann <daniel@...earbox.net>
To:     davem@...emloft.net
Cc:     daniel@...earbox.net, ast@...nel.org, netdev@...r.kernel.org
Subject: pull-request: bpf 2018-05-18

Hi David,

The following pull-request contains BPF updates for your *net* tree.

The main changes are:

1) Fix two bugs in sockmap, a use after free in sockmap's error path
   from sock_map_ctx_update_elem() where we mistakenly drop a reference
   we didn't take prior to that, and in the same function fix a race
   in bpf_prog_inc_not_zero() where we didn't use the progs from prior
   READ_ONCE(), from John.

2) Reject program expansions once we figure out that their jump target
   which crosses patchlet boundaries could otherwise get truncated in
   insn->off space, from Daniel.

3) Check the return value of fopen() in BPF selftest's test_verifier
   where we determine whether unpriv BPF is disabled, and iff we do
   fail there then just assume it is disabled. This fixes a segfault
   when used with older kernels, from Jesper.

Please consider pulling these changes from:

  git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf.git

When this gets later merged into net-next there are a two trivial
BPF conflicts to resolve:

In kernel/bpf/sockmap.c the bpf_prog_inc_not_zero() cases must
use verdict, parse and tx_msg as their arguments as opposed to
the buggy old version where progs->bpf_{verdict,parse,tx_msg}
were used as passed args.

In tools/lib/bpf/libbpf.c use the hunk from net-next with the
__bpf_object__open() + IS_ERR(obj) test combination. Thus, net-next
code only is sufficient here.

Thanks a lot!

----------------------------------------------------------------

The following changes since commit 02f99df1875c11330cd0be69a40fa8ccd14749b2:

  erspan: fix invalid erspan version. (2018-05-17 15:48:49 -0400)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf.git 

for you to fetch changes up to 050fad7c4534c13c8eb1d9c2ba66012e014773cb:

  bpf: fix truncated jump targets on heavy expansions (2018-05-17 16:05:35 -0700)

----------------------------------------------------------------
Daniel Borkmann (1):
      bpf: fix truncated jump targets on heavy expansions

Jesper Dangaard Brouer (1):
      selftests/bpf: check return value of fopen in test_verifier.c

John Fastabend (2):
      bpf: sockmap update rollback on error can incorrectly dec prog refcnt
      bpf: parse and verdict prog attach may race with bpf map update

 kernel/bpf/core.c                           | 100 +++++++++++++++++++++-------
 kernel/bpf/sockmap.c                        |  18 ++---
 net/core/filter.c                           |  11 ++-
 tools/testing/selftests/bpf/test_verifier.c |   5 ++
 4 files changed, 98 insertions(+), 36 deletions(-)