bpf: fix incorrect sign extension in check_alu_op()
authorDaniel Borkmann <daniel@iogearbox.net>
Fri, 22 Dec 2017 15:23:05 +0000 (16:23 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 25 Dec 2017 13:26:32 +0000 (14:26 +0100)
commit6e12ea4fb45ca86cdd7425276b6993455fee947a
tree6226760084a3be2a60260c60d46a509213c6e8e1
parent4d54f7df5131d67f653f674003ec5f52c9818b53
bpf: fix incorrect sign extension in check_alu_op()

From: Jann Horn <jannh@google.com>

[ Upstream commit 95a762e2c8c942780948091f8f2a4f32fce1ac6f ]

Distinguish between
BPF_ALU64|BPF_MOV|BPF_K (load 32-bit immediate, sign-extended to 64-bit)
and BPF_ALU|BPF_MOV|BPF_K (load 32-bit immediate, zero-padded to 64-bit);
only perform sign extension in the first case.

Starting with v4.14, this is exploitable by unprivileged users as long as
the unprivileged_bpf_disabled sysctl isn't set.

Debian assigned CVE-2017-16995 for this issue.

v3:
 - add CVE number (Ben Hutchings)

Fixes: 484611357c19 ("bpf: allow access into map value arrays")
Signed-off-by: Jann Horn <jannh@google.com>
Acked-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
kernel/bpf/verifier.c