[PATCH net-next] bpf: test for AND edge cases

[PATCH net-next] bpf: test for AND edge cases

[Josef Bacik]

These two tests are based on the work done for f23cc643f9ba.  The first test is
just a basic one to make sure we don't allow AND'ing negative values, even if it
would result in a valid index for the array.  The second is a cleaned up version
of the original testcase provided by Jann Horn that resulted in the commit.

Signed-off-by: Josef Bacik <jbacik@fb.com>
---
 tools/testing/selftests/bpf/test_verifier.c | 55 +++++++++++++++++++++++++++++
 1 file changed, 55 insertions(+)

diff --git a/tools/testing/selftests/bpf/test_verifier.c b/tools/testing/selftests/bpf/test_verifier.c
index 853d7e4..44404f1 100644
--- a/tools/testing/selftests/bpf/test_verifier.c
+++ b/tools/testing/selftests/bpf/test_verifier.c
@@ -2905,6 +2905,61 @@ static struct bpf_test tests[] = {
 		.result = REJECT,
 		.errstr = "invalid bpf_context access",
 	},
+	{
+		"invalid and of negative number",
+		.insns = {
+			BPF_ST_MEM(BPF_DW, BPF_REG_10, -8, 0),
+			BPF_MOV64_REG(BPF_REG_2, BPF_REG_10),
+			BPF_ALU64_IMM(BPF_ADD, BPF_REG_2, -8),
+			BPF_LD_MAP_FD(BPF_REG_1, 0),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_map_lookup_elem),
+			BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 4),
+			BPF_MOV64_IMM(BPF_REG_1, 6),
+			BPF_ALU64_IMM(BPF_AND, BPF_REG_1, -4),
+			BPF_ALU64_IMM(BPF_LSH, BPF_REG_1, 2),
+			BPF_ALU64_REG(BPF_ADD, BPF_REG_0, BPF_REG_1),
+			BPF_ST_MEM(BPF_DW, BPF_REG_0, 0,
+				   offsetof(struct test_val, foo)),
+			BPF_EXIT_INSN(),
+		},
+		.fixup_map2 = { 3 },
+		.errstr_unpriv = "R0 pointer arithmetic prohibited",
+		.errstr = "R0 min value is negative, either use unsigned index or do a if (index >=0) check.",
+		.result = REJECT,
+		.result_unpriv = REJECT,
+	},
+	{
+		"invalid range check",
+		.insns = {
+			BPF_ST_MEM(BPF_DW, BPF_REG_10, -8, 0),
+			BPF_MOV64_REG(BPF_REG_2, BPF_REG_10),
+			BPF_ALU64_IMM(BPF_ADD, BPF_REG_2, -8),
+			BPF_LD_MAP_FD(BPF_REG_1, 0),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_map_lookup_elem),
+			BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 12),
+			BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, 0),
+			BPF_MOV64_IMM(BPF_REG_9, 1),
+			BPF_ALU32_IMM(BPF_MOD, BPF_REG_1, 2),
+			BPF_ALU32_IMM(BPF_ADD, BPF_REG_1, 1),
+			BPF_ALU32_REG(BPF_AND, BPF_REG_9, BPF_REG_1),
+			BPF_ALU32_IMM(BPF_ADD, BPF_REG_9, 1),
+			BPF_ALU32_IMM(BPF_RSH, BPF_REG_9, 1),
+			BPF_MOV32_IMM(BPF_REG_3, 1),
+			BPF_ALU32_REG(BPF_SUB, BPF_REG_3, BPF_REG_9),
+			BPF_ALU32_IMM(BPF_MUL, BPF_REG_3, 0x10000000),
+			BPF_ALU64_REG(BPF_ADD, BPF_REG_0, BPF_REG_3),
+			BPF_STX_MEM(BPF_W, BPF_REG_0, BPF_REG_3, 0),
+			BPF_MOV64_REG(BPF_REG_0, 0),
+			BPF_EXIT_INSN(),
+		},
+		.fixup_map2 = { 3 },
+		.errstr_unpriv = "R0 pointer arithmetic prohibited",
+		.errstr = "R0 min value is negative, either use unsigned index or do a if (index >=0) check.",
+		.result = REJECT,
+		.result_unpriv = REJECT,
+	}
 };
 
 static int probe_filter_length(const struct bpf_insn *fp)
-- 
2.7.4

Re: [PATCH net-next] bpf: test for AND edge cases

[Alexei Starovoitov]

On 2/2/17 9:00 AM, Josef Bacik wrote:
> These two tests are based on the work done for f23cc643f9ba.  The first test is
> just a basic one to make sure we don't allow AND'ing negative values, even if it
> would result in a valid index for the array.  The second is a cleaned up version
> of the original testcase provided by Jann Horn that resulted in the commit.
>
> Signed-off-by: Josef Bacik <jbacik@fb.com>

Thanks for the tests! Much appreciated.

> diff --git a/tools/testing/selftests/bpf/test_verifier.c b/tools/testing/selftests/bpf/test_verifier.c
> index 853d7e4..44404f1 100644
> --- a/tools/testing/selftests/bpf/test_verifier.c
> +++ b/tools/testing/selftests/bpf/test_verifier.c
> @@ -2905,6 +2905,61 @@ static struct bpf_test tests[] = {
>   		.result = REJECT,
>   		.errstr = "invalid bpf_context access",
>   	},
> +	{
> +		"invalid and of negative number",
> +		.insns = {
> +			BPF_ST_MEM(BPF_DW, BPF_REG_10, -8, 0),
> +			BPF_MOV64_REG(BPF_REG_2, BPF_REG_10),
> +			BPF_ALU64_IMM(BPF_ADD, BPF_REG_2, -8),
> +			BPF_LD_MAP_FD(BPF_REG_1, 0),
> +			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
> +				     BPF_FUNC_map_lookup_elem),
> +			BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 4),
> +			BPF_MOV64_IMM(BPF_REG_1, 6),
> +			BPF_ALU64_IMM(BPF_AND, BPF_REG_1, -4),
> +			BPF_ALU64_IMM(BPF_LSH, BPF_REG_1, 2),
> +			BPF_ALU64_REG(BPF_ADD, BPF_REG_0, BPF_REG_1),
> +			BPF_ST_MEM(BPF_DW, BPF_REG_0, 0,
> +				   offsetof(struct test_val, foo)),
> +			BPF_EXIT_INSN(),
> +		},
> +		.fixup_map2 = { 3 },
> +		.errstr_unpriv = "R0 pointer arithmetic prohibited",
> +		.errstr = "R0 min value is negative, either use unsigned index or do a if (index >=0) check.",

the errstr doesn't have to compare the whole string. In case we find
typos or adjust the hint message, we'd need to change the test as well,
but I see it's being used as-is in other tests already, so we'll
fix all of them at once when time comes.

Acked-by: Alexei Starovoitov <ast@kernel.org>

Re: [PATCH net-next] bpf: test for AND edge cases

[Daniel Borkmann]

On 02/02/2017 06:00 PM, Josef Bacik wrote:
> These two tests are based on the work done for f23cc643f9ba.  The first test is
> just a basic one to make sure we don't allow AND'ing negative values, even if it
> would result in a valid index for the array.  The second is a cleaned up version
> of the original testcase provided by Jann Horn that resulted in the commit.
>
> Signed-off-by: Josef Bacik <jbacik@fb.com>

Thanks for following up!

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

Re: [PATCH net-next] bpf: test for AND edge cases

[David Miller]

From: Josef Bacik <jbacik@fb.com>
Date: Thu, 2 Feb 2017 12:00:38 -0500

> These two tests are based on the work done for f23cc643f9ba.  The first test is
> just a basic one to make sure we don't allow AND'ing negative values, even if it
> would result in a valid index for the array.  The second is a cleaned up version
> of the original testcase provided by Jann Horn that resulted in the commit.
> 
> Signed-off-by: Josef Bacik <jbacik@fb.com>

This doesn't apply cleanly to net-next, please respin.

Re: [PATCH net-next] bpf: test for AND edge cases

[Josef Bacik]

On Fri, 2017-02-03 at 16:03 -0500, David Miller wrote:
> From: Josef Bacik <jbacik@fb.com>
> Date: Thu, 2 Feb 2017 12:00:38 -0500
> 
> > 
> > These two tests are based on the work done for f23cc643f9ba.  The
> > first test is
> > just a basic one to make sure we don't allow AND'ing negative
> > values, even if it
> > would result in a valid index for the array.  The second is a
> > cleaned up version
> > of the original testcase provided by Jann Horn that resulted in the
> > commit.
> > 
> > Signed-off-by: Josef Bacik <jbacik@fb.com>
> This doesn't apply cleanly to net-next, please respin.

Ugh sorry did it on the wrong branch, I'll send an updated one shortly.
 Thanks,

Josef