| Message ID | 1562837513-745-1-git-send-email-p.pisati@gmail.com |
|---|---|
| Headers | show |
| Series | Fold checksum at the end of bpf_csum_diff and fix | expand |
On Thu, Jul 11, 2019 at 2:32 AM Paolo Pisati <p.pisati@gmail.com> wrote: > > From: Paolo Pisati <paolo.pisati@canonical.com> > > After applying patch 0001, all checksum implementations i could test (x86-64, arm64 and > arm), now agree on the return value. > > Patch 0002 fix the expected return value for test #13: i did the calculation manually, > and it correspond. > > Unfortunately, after applying patch 0001, other test cases now fail in > test_verifier: > > $ sudo ./tools/testing/selftests/bpf/test_verifier > ... > #417/p helper access to variable memory: size = 0 allowed on NULL (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 > #419/p helper access to variable memory: size = 0 allowed on != NULL stack pointer (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 > #423/p helper access to variable memory: size possible = 0 allowed on != NULL packet pointer (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 I'm not entirely sure this fix is correct, given these failures, to be honest. Let's wait for someone who understands intended semantics for bpf_csum_diff, before changing returned value so drastically. But in any case, fixes for these test failures should be in your patch series as well. > ... > Summary: 1500 PASSED, 0 SKIPPED, 3 FAILED > > And there are probably other fallouts in other selftests - someone familiar > should take a look before applying these patches. > > Paolo Pisati (2): > bpf: bpf_csum_diff: fold the checksum before returning the > value > bpf, selftest: fix checksum value for test #13 > > net/core/filter.c | 2 +- > tools/testing/selftests/bpf/verifier/array_access.c | 2 +- > 2 files changed, 2 insertions(+), 2 deletions(-) > > -- > 2.17.1 >
On 07/12/2019 01:50 AM, Andrii Nakryiko wrote: > On Thu, Jul 11, 2019 at 2:32 AM Paolo Pisati <p.pisati@gmail.com> wrote: >> From: Paolo Pisati <paolo.pisati@canonical.com> >> >> After applying patch 0001, all checksum implementations i could test (x86-64, arm64 and >> arm), now agree on the return value. >> >> Patch 0002 fix the expected return value for test #13: i did the calculation manually, >> and it correspond. >> >> Unfortunately, after applying patch 0001, other test cases now fail in >> test_verifier: Thanks for catching, sigh. :/ >> $ sudo ./tools/testing/selftests/bpf/test_verifier >> ... >> #417/p helper access to variable memory: size = 0 allowed on NULL (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 >> #419/p helper access to variable memory: size = 0 allowed on != NULL stack pointer (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 >> #423/p helper access to variable memory: size possible = 0 allowed on != NULL packet pointer (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 > > I'm not entirely sure this fix is correct, given these failures, to be honest. > > Let's wait for someone who understands intended semantics for > bpf_csum_diff, before changing returned value so drastically. > > But in any case, fixes for these test failures should be in your patch > series as well. Your change would actually break applications. The bpf_csum_diff() helper is heavily used with cascading so one result can be fed into another bpf_csum_diff() call as seed. Quick test on x86-64: static int __init foo(void) { u8 data[32 * sizeof(u32)]; u32 res1, res2, res3; int i; prandom_bytes(data, sizeof(data)); res1 = csum_fold(csum_partial(data, sizeof(data), 0)); for (i = sizeof(u32); i < sizeof(data); i += sizeof(u32)) { res2 = csum_fold(csum_partial(data, i, 0)); res2 = csum_fold(csum_partial(data+i, sizeof(data)-i, res2)); res3 = csum_partial(data, i, 0); res3 = csum_fold(csum_partial(data+i, sizeof(data)-i, res3)); printk("%8d: [%4x (reference), %4x (unfolded), %4x (folded)]\n", i, res1, res3, res2); } return -1; } Gives for all three: [19113.233942] 4: [6b70 (reference), 6b70 (unfolded), 223d (folded)] [19113.233943] 8: [6b70 (reference), 6b70 (unfolded), a812 (folded)] [19113.233943] 12: [6b70 (reference), 6b70 (unfolded), 1c26 (folded)] [19113.233944] 16: [6b70 (reference), 6b70 (unfolded), 4f76 (folded)] [19113.233944] 20: [6b70 (reference), 6b70 (unfolded), 2801 (folded)] [19113.233945] 24: [6b70 (reference), 6b70 (unfolded), b63 (folded)] [19113.233945] 28: [6b70 (reference), 6b70 (unfolded), 2fe0 (folded)] [19113.233946] 32: [6b70 (reference), 6b70 (unfolded), 18a2 (folded)] [19113.233946] 36: [6b70 (reference), 6b70 (unfolded), 2597 (folded)] [19113.233947] 40: [6b70 (reference), 6b70 (unfolded), 2f8e (folded)] [19113.233947] 44: [6b70 (reference), 6b70 (unfolded), b8af (folded)] [19113.233948] 48: [6b70 (reference), 6b70 (unfolded), fb8b (folded)] [19113.233948] 52: [6b70 (reference), 6b70 (unfolded), e9c0 (folded)] [19113.233949] 56: [6b70 (reference), 6b70 (unfolded), 6af1 (folded)] [19113.233949] 60: [6b70 (reference), 6b70 (unfolded), d7f4 (folded)] [19113.233949] 64: [6b70 (reference), 6b70 (unfolded), 8bc6 (folded)] [19113.233950] 68: [6b70 (reference), 6b70 (unfolded), 8718 (folded)] [19113.233950] 72: [6b70 (reference), 6b70 (unfolded), 27d8 (folded)] [19113.233951] 76: [6b70 (reference), 6b70 (unfolded), a2db (folded)] [19113.233952] 80: [6b70 (reference), 6b70 (unfolded), 3fd (folded)] [19113.233952] 84: [6b70 (reference), 6b70 (unfolded), 4be5 (folded)] [19113.233952] 88: [6b70 (reference), 6b70 (unfolded), 41ad (folded)] [19113.233953] 92: [6b70 (reference), 6b70 (unfolded), ca9b (folded)] [19113.233953] 96: [6b70 (reference), 6b70 (unfolded), f8ec (folded)] [19113.233954] 100: [6b70 (reference), 6b70 (unfolded), 5451 (folded)] [19113.233954] 104: [6b70 (reference), 6b70 (unfolded), 763 (folded)] [19113.233955] 108: [6b70 (reference), 6b70 (unfolded), e37c (folded)] [19113.233955] 112: [6b70 (reference), 6b70 (unfolded), 4ee6 (folded)] [19113.233956] 116: [6b70 (reference), 6b70 (unfolded), 4f73 (folded)] [19113.233956] 120: [6b70 (reference), 6b70 (unfolded), 1cfd (folded)] [19113.233957] 124: [6b70 (reference), 6b70 (unfolded), 7d1a (folded)] I'll take a look next week wrt fixing this uniformly for all archs. Thanks, Daniel
From: Paolo Pisati <paolo.pisati@canonical.com> After applying patch 0001, all checksum implementations i could test (x86-64, arm64 and arm), now agree on the return value. Patch 0002 fix the expected return value for test #13: i did the calculation manually, and it correspond. Unfortunately, after applying patch 0001, other test cases now fail in test_verifier: $ sudo ./tools/testing/selftests/bpf/test_verifier ... #417/p helper access to variable memory: size = 0 allowed on NULL (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 #419/p helper access to variable memory: size = 0 allowed on != NULL stack pointer (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 #423/p helper access to variable memory: size possible = 0 allowed on != NULL packet pointer (ARG_PTR_TO_MEM_OR_NULL) FAIL retval 65535 != 0 ... Summary: 1500 PASSED, 0 SKIPPED, 3 FAILED And there are probably other fallouts in other selftests - someone familiar should take a look before applying these patches. Paolo Pisati (2): bpf: bpf_csum_diff: fold the checksum before returning the value bpf, selftest: fix checksum value for test #13 net/core/filter.c | 2 +- tools/testing/selftests/bpf/verifier/array_access.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-)