From: Xu Kuohai xukuohai@huawei.com
The two infinite loop bound check cases added by commit 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") take a long time to execute but don't add much value.
Remove them to reduce run time of test_verifier.
Fixes: 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") Signed-off-by: Xu Kuohai xukuohai@huawei.com --- tools/testing/selftests/bpf/verifier/bounds.c | 50 ------------------- 1 file changed, 50 deletions(-)
diff --git a/tools/testing/selftests/bpf/verifier/bounds.c b/tools/testing/selftests/bpf/verifier/bounds.c index 74b1917d4208..515a8222f08f 100644 --- a/tools/testing/selftests/bpf/verifier/bounds.c +++ b/tools/testing/selftests/bpf/verifier/bounds.c @@ -777,31 +777,6 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{ - "bound check with JMP_JSLT for crossing 64-bit signed boundary", - .insns = { - BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)), - BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)), - BPF_MOV64_REG(BPF_REG_1, BPF_REG_2), - BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1), - BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 8), - - BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0), - BPF_LD_IMM64(BPF_REG_0, 0x7fffffffffffff10), - BPF_ALU64_REG(BPF_ADD, BPF_REG_1, BPF_REG_0), - - BPF_LD_IMM64(BPF_REG_0, 0x8000000000000000), - BPF_ALU64_IMM(BPF_ADD, BPF_REG_0, 1), - /* r1 signed range is [S64_MIN, S64_MAX] */ - BPF_JMP_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2), - - BPF_MOV64_IMM(BPF_REG_0, 0), - BPF_EXIT_INSN(), - }, - .errstr = "BPF program is too large", - .result = REJECT, - .prog_type = BPF_PROG_TYPE_XDP, -}, { "bound check for loop upper bound greater than U32_MAX", .insns = { @@ -849,28 +824,3 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{ - "bound check with JMP32_JSLT for crossing 32-bit signed boundary", - .insns = { - BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)), - BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)), - BPF_MOV64_REG(BPF_REG_1, BPF_REG_2), - BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1), - BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 6), - - BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0), - BPF_MOV32_IMM(BPF_REG_0, 0x7fffff10), - BPF_ALU32_REG(BPF_ADD, BPF_REG_1, BPF_REG_0), - - BPF_MOV32_IMM(BPF_REG_0, 0x80000000), - BPF_ALU32_IMM(BPF_ADD, BPF_REG_0, 1), - /* r1 signed range is [S32_MIN, S32_MAX] */ - BPF_JMP32_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2), - - BPF_MOV64_IMM(BPF_REG_0, 0), - BPF_EXIT_INSN(), - }, - .errstr = "BPF program is too large", - .result = REJECT, - .prog_type = BPF_PROG_TYPE_XDP, -},
On Sun, Mar 26, 2023 at 7:45 PM Xu Kuohai xukuohai@huaweicloud.com wrote:
From: Xu Kuohai xukuohai@huawei.com
The two infinite loop bound check cases added by commit 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") take a long time to execute but don't add much value.
Remove them to reduce run time of test_verifier.
Summary: 2042 PASSED, 0 SKIPPED, 1 FAILED
real 0m4.780s user 0m0.458s sys 0m3.871s
5 seconds isn't such a long time, especially when we compare it to test_progs (even with parallelization).
Fixes: 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") Signed-off-by: Xu Kuohai xukuohai@huawei.com
tools/testing/selftests/bpf/verifier/bounds.c | 50 ------------------- 1 file changed, 50 deletions(-)
diff --git a/tools/testing/selftests/bpf/verifier/bounds.c b/tools/testing/selftests/bpf/verifier/bounds.c index 74b1917d4208..515a8222f08f 100644 --- a/tools/testing/selftests/bpf/verifier/bounds.c +++ b/tools/testing/selftests/bpf/verifier/bounds.c @@ -777,31 +777,6 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{
"bound check with JMP_JSLT for crossing 64-bit signed boundary",.insns = {BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)),BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)),BPF_MOV64_REG(BPF_REG_1, BPF_REG_2),BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 8),BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0),BPF_LD_IMM64(BPF_REG_0, 0x7fffffffffffff10),BPF_ALU64_REG(BPF_ADD, BPF_REG_1, BPF_REG_0),BPF_LD_IMM64(BPF_REG_0, 0x8000000000000000),BPF_ALU64_IMM(BPF_ADD, BPF_REG_0, 1),/* r1 signed range is [S64_MIN, S64_MAX] */BPF_JMP_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2),BPF_MOV64_IMM(BPF_REG_0, 0),BPF_EXIT_INSN(),},.errstr = "BPF program is too large",.result = REJECT,.prog_type = BPF_PROG_TYPE_XDP,-}, { "bound check for loop upper bound greater than U32_MAX", .insns = { @@ -849,28 +824,3 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{
"bound check with JMP32_JSLT for crossing 32-bit signed boundary",.insns = {BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)),BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)),BPF_MOV64_REG(BPF_REG_1, BPF_REG_2),BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 6),BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0),BPF_MOV32_IMM(BPF_REG_0, 0x7fffff10),BPF_ALU32_REG(BPF_ADD, BPF_REG_1, BPF_REG_0),BPF_MOV32_IMM(BPF_REG_0, 0x80000000),BPF_ALU32_IMM(BPF_ADD, BPF_REG_0, 1),/* r1 signed range is [S32_MIN, S32_MAX] */BPF_JMP32_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2),BPF_MOV64_IMM(BPF_REG_0, 0),BPF_EXIT_INSN(),},.errstr = "BPF program is too large",.result = REJECT,.prog_type = BPF_PROG_TYPE_XDP,-},
2.30.2
On 3/27/2023 11:20 AM, Andrii Nakryiko wrote:
On Sun, Mar 26, 2023 at 7:45 PM Xu Kuohai xukuohai@huaweicloud.com wrote:
From: Xu Kuohai xukuohai@huawei.com
The two infinite loop bound check cases added by commit 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") take a long time to execute but don't add much value.
Remove them to reduce run time of test_verifier.
Summary: 2042 PASSED, 0 SKIPPED, 1 FAILED
real 0m4.780s user 0m0.458s sys 0m3.871s
5 seconds isn't such a long time, especially when we compare it to test_progs (even with parallelization).
Well, I actually don't know if it is "long time".
This patch was sent to address Alexei's concern about the run time of test_verifier in mail [1].
[1] https://lore.kernel.org/bpf/20230322213056.2470-1-daniel@iogearbox.net/T/#mb...
Fixes: 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") Signed-off-by: Xu Kuohai xukuohai@huawei.com
tools/testing/selftests/bpf/verifier/bounds.c | 50 ------------------- 1 file changed, 50 deletions(-)
diff --git a/tools/testing/selftests/bpf/verifier/bounds.c b/tools/testing/selftests/bpf/verifier/bounds.c index 74b1917d4208..515a8222f08f 100644 --- a/tools/testing/selftests/bpf/verifier/bounds.c +++ b/tools/testing/selftests/bpf/verifier/bounds.c @@ -777,31 +777,6 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{
"bound check with JMP_JSLT for crossing 64-bit signed boundary",.insns = {BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)),BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)),BPF_MOV64_REG(BPF_REG_1, BPF_REG_2),BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 8),BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0),BPF_LD_IMM64(BPF_REG_0, 0x7fffffffffffff10),BPF_ALU64_REG(BPF_ADD, BPF_REG_1, BPF_REG_0),BPF_LD_IMM64(BPF_REG_0, 0x8000000000000000),BPF_ALU64_IMM(BPF_ADD, BPF_REG_0, 1),/* r1 signed range is [S64_MIN, S64_MAX] */BPF_JMP_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2),BPF_MOV64_IMM(BPF_REG_0, 0),BPF_EXIT_INSN(),},.errstr = "BPF program is too large",.result = REJECT,.prog_type = BPF_PROG_TYPE_XDP,-}, { "bound check for loop upper bound greater than U32_MAX", .insns = { @@ -849,28 +824,3 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{
"bound check with JMP32_JSLT for crossing 32-bit signed boundary",.insns = {BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)),BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)),BPF_MOV64_REG(BPF_REG_1, BPF_REG_2),BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 6),BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0),BPF_MOV32_IMM(BPF_REG_0, 0x7fffff10),BPF_ALU32_REG(BPF_ADD, BPF_REG_1, BPF_REG_0),BPF_MOV32_IMM(BPF_REG_0, 0x80000000),BPF_ALU32_IMM(BPF_ADD, BPF_REG_0, 1),/* r1 signed range is [S32_MIN, S32_MAX] */BPF_JMP32_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2),BPF_MOV64_IMM(BPF_REG_0, 0),BPF_EXIT_INSN(),},.errstr = "BPF program is too large",.result = REJECT,.prog_type = BPF_PROG_TYPE_XDP,-},
2.30.2
.
On Sun, Mar 26, 2023 at 11:21 PM Xu Kuohai xukuohai@huaweicloud.com wrote:
On 3/27/2023 11:20 AM, Andrii Nakryiko wrote:
On Sun, Mar 26, 2023 at 7:45 PM Xu Kuohai xukuohai@huaweicloud.com wrote:
From: Xu Kuohai xukuohai@huawei.com
The two infinite loop bound check cases added by commit 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") take a long time to execute but don't add much value.
Remove them to reduce run time of test_verifier.
Summary: 2042 PASSED, 0 SKIPPED, 1 FAILED
real 0m4.780s user 0m0.458s sys 0m3.871s
5 seconds isn't such a long time, especially when we compare it to test_progs (even with parallelization).
Well, I actually don't know if it is "long time".
This patch was sent to address Alexei's concern about the run time of test_verifier in mail [1].
These infinite loops don't add much value to the actual test. Please rewrite them without infinite loops.
Alexei asked to improve the test, not to just remove it, if I understand correctly. If the test is there, presumably it's useful. If it can be implemented better, let's do that. Just removing the test seems like a wrong move.
[1] https://lore.kernel.org/bpf/20230322213056.2470-1-daniel@iogearbox.net/T/#mb...
Fixes: 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") Signed-off-by: Xu Kuohai xukuohai@huawei.com
tools/testing/selftests/bpf/verifier/bounds.c | 50 ------------------- 1 file changed, 50 deletions(-)
diff --git a/tools/testing/selftests/bpf/verifier/bounds.c b/tools/testing/selftests/bpf/verifier/bounds.c index 74b1917d4208..515a8222f08f 100644 --- a/tools/testing/selftests/bpf/verifier/bounds.c +++ b/tools/testing/selftests/bpf/verifier/bounds.c @@ -777,31 +777,6 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{
"bound check with JMP_JSLT for crossing 64-bit signed boundary",.insns = {BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)),BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)),BPF_MOV64_REG(BPF_REG_1, BPF_REG_2),BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 8),BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0),BPF_LD_IMM64(BPF_REG_0, 0x7fffffffffffff10),BPF_ALU64_REG(BPF_ADD, BPF_REG_1, BPF_REG_0),BPF_LD_IMM64(BPF_REG_0, 0x8000000000000000),BPF_ALU64_IMM(BPF_ADD, BPF_REG_0, 1),/* r1 signed range is [S64_MIN, S64_MAX] */BPF_JMP_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2),BPF_MOV64_IMM(BPF_REG_0, 0),BPF_EXIT_INSN(),},.errstr = "BPF program is too large",.result = REJECT,.prog_type = BPF_PROG_TYPE_XDP,-}, { "bound check for loop upper bound greater than U32_MAX", .insns = { @@ -849,28 +824,3 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{
"bound check with JMP32_JSLT for crossing 32-bit signed boundary",.insns = {BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)),BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)),BPF_MOV64_REG(BPF_REG_1, BPF_REG_2),BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 6),BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0),BPF_MOV32_IMM(BPF_REG_0, 0x7fffff10),BPF_ALU32_REG(BPF_ADD, BPF_REG_1, BPF_REG_0),BPF_MOV32_IMM(BPF_REG_0, 0x80000000),BPF_ALU32_IMM(BPF_ADD, BPF_REG_0, 1),/* r1 signed range is [S32_MIN, S32_MAX] */BPF_JMP32_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2),BPF_MOV64_IMM(BPF_REG_0, 0),BPF_EXIT_INSN(),},.errstr = "BPF program is too large",.result = REJECT,.prog_type = BPF_PROG_TYPE_XDP,-},
2.30.2
.
On 3/28/2023 2:55 AM, Andrii Nakryiko wrote:
On Sun, Mar 26, 2023 at 11:21 PM Xu Kuohai xukuohai@huaweicloud.com wrote:
On 3/27/2023 11:20 AM, Andrii Nakryiko wrote:
On Sun, Mar 26, 2023 at 7:45 PM Xu Kuohai xukuohai@huaweicloud.com wrote:
From: Xu Kuohai xukuohai@huawei.com
The two infinite loop bound check cases added by commit 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") take a long time to execute but don't add much value.
Remove them to reduce run time of test_verifier.
Summary: 2042 PASSED, 0 SKIPPED, 1 FAILED
real 0m4.780s user 0m0.458s sys 0m3.871s
5 seconds isn't such a long time, especially when we compare it to test_progs (even with parallelization).
Well, I actually don't know if it is "long time".
This patch was sent to address Alexei's concern about the run time of test_verifier in mail [1].
These infinite loops don't add much value to the actual test. Please rewrite them without infinite loops.
Alexei asked to improve the test, not to just remove it, if I understand correctly. If the test is there, presumably it's useful. If it can be implemented better, let's do that. Just removing the test seems like a wrong move.
OK. I added upper bounds to the two loops to make them finite.
[1] https://lore.kernel.org/bpf/20230322213056.2470-1-daniel@iogearbox.net/T/#mb...
Fixes: 1a3148fc171f ("selftests/bpf: Check when bounds are not in the 32-bit range") Signed-off-by: Xu Kuohai xukuohai@huawei.com
tools/testing/selftests/bpf/verifier/bounds.c | 50 ------------------- 1 file changed, 50 deletions(-)
diff --git a/tools/testing/selftests/bpf/verifier/bounds.c b/tools/testing/selftests/bpf/verifier/bounds.c index 74b1917d4208..515a8222f08f 100644 --- a/tools/testing/selftests/bpf/verifier/bounds.c +++ b/tools/testing/selftests/bpf/verifier/bounds.c @@ -777,31 +777,6 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{
"bound check with JMP_JSLT for crossing 64-bit signed boundary",.insns = {BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)),BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)),BPF_MOV64_REG(BPF_REG_1, BPF_REG_2),BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 8),BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0),BPF_LD_IMM64(BPF_REG_0, 0x7fffffffffffff10),BPF_ALU64_REG(BPF_ADD, BPF_REG_1, BPF_REG_0),BPF_LD_IMM64(BPF_REG_0, 0x8000000000000000),BPF_ALU64_IMM(BPF_ADD, BPF_REG_0, 1),/* r1 signed range is [S64_MIN, S64_MAX] */BPF_JMP_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2),BPF_MOV64_IMM(BPF_REG_0, 0),BPF_EXIT_INSN(),},.errstr = "BPF program is too large",.result = REJECT,.prog_type = BPF_PROG_TYPE_XDP,-}, { "bound check for loop upper bound greater than U32_MAX", .insns = { @@ -849,28 +824,3 @@ .result = ACCEPT, .prog_type = BPF_PROG_TYPE_XDP, }, -{
"bound check with JMP32_JSLT for crossing 32-bit signed boundary",.insns = {BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, offsetof(struct xdp_md, data)),BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1, offsetof(struct xdp_md, data_end)),BPF_MOV64_REG(BPF_REG_1, BPF_REG_2),BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),BPF_JMP_REG(BPF_JGT, BPF_REG_1, BPF_REG_3, 6),BPF_LDX_MEM(BPF_B, BPF_REG_1, BPF_REG_2, 0),BPF_MOV32_IMM(BPF_REG_0, 0x7fffff10),BPF_ALU32_REG(BPF_ADD, BPF_REG_1, BPF_REG_0),BPF_MOV32_IMM(BPF_REG_0, 0x80000000),BPF_ALU32_IMM(BPF_ADD, BPF_REG_0, 1),/* r1 signed range is [S32_MIN, S32_MAX] */BPF_JMP32_REG(BPF_JSLT, BPF_REG_0, BPF_REG_1, -2),BPF_MOV64_IMM(BPF_REG_0, 0),BPF_EXIT_INSN(),},.errstr = "BPF program is too large",.result = REJECT,.prog_type = BPF_PROG_TYPE_XDP,-},
2.30.2
.
.
linux-kselftest-mirror@lists.linaro.org
-
Andrii Nakryiko -
Xu Kuohai