[PATCH 2/4][A] bpf: fix incorrect sign extension in check_alu_op()

[Seth Forshee]

From: Jann Horn <jannh at 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 at google.com>
Acked-by: Edward Cree <ecree at solarflare.com>
Signed-off-by: Alexei Starovoitov <ast at kernel.org>
Signed-off-by: Daniel Borkmann <daniel at iogearbox.net>
CVE-2017-16995
Signed-off-by: Seth Forshee <seth.forshee at canonical.com>
---
 kernel/bpf/verifier.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 4321625fe32a..cdfa07a4ef27 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -2048,12 +2048,19 @@ static int check_alu_op(struct bpf_verifier_env *env, struct bpf_insn *insn)
 			/* case: R = imm
 			 * remember the value we stored into this reg
 			 */
+			u64 imm;
+
+			if (BPF_CLASS(insn->code) == BPF_ALU64)
+				imm = insn->imm;
+			else
+				imm = (u32)insn->imm;
+
 			regs[insn->dst_reg].type = CONST_IMM;
-			regs[insn->dst_reg].imm = insn->imm;
+			regs[insn->dst_reg].imm = imm;
 			regs[insn->dst_reg].id = 0;
-			regs[insn->dst_reg].max_value = insn->imm;
-			regs[insn->dst_reg].min_value = insn->imm;
-			regs[insn->dst_reg].min_align = calc_align(insn->imm);
+			regs[insn->dst_reg].max_value = imm;
+			regs[insn->dst_reg].min_value = imm;
+			regs[insn->dst_reg].min_align = calc_align(imm);
 			regs[insn->dst_reg].value_from_signed = false;
 		}
 
-- 
2.14.1