1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
|
==================
BPF Selftest Notes
==================
General instructions on running selftests can be found in
`Documentation/bpf/bpf_devel_QA.rst`__.
__ /Documentation/bpf/bpf_devel_QA.rst#q-how-to-run-bpf-selftests
Additional information about selftest failures are
documented here.
profiler[23] test failures with clang/llvm <12.0.0
==================================================
With clang/llvm <12.0.0, the profiler[23] test may fail.
The symptom looks like
.. code-block:: c
// r9 is a pointer to map_value
// r7 is a scalar
17: bf 96 00 00 00 00 00 00 r6 = r9
18: 0f 76 00 00 00 00 00 00 r6 += r7
math between map_value pointer and register with unbounded min value is not allowed
// the instructions below will not be seen in the verifier log
19: a5 07 01 00 01 01 00 00 if r7 < 257 goto +1
20: bf 96 00 00 00 00 00 00 r6 = r9
// r6 is used here
The verifier will reject such code with above error.
At insn 18 the r7 is indeed unbounded. The later insn 19 checks the bounds and
the insn 20 undoes map_value addition. It is currently impossible for the
verifier to understand such speculative pointer arithmetic.
Hence `this patch`__ addresses it on the compiler side. It was committed on llvm 12.
__ https://reviews.llvm.org/D85570
The corresponding C code
.. code-block:: c
for (int i = 0; i < MAX_CGROUPS_PATH_DEPTH; i++) {
filepart_length = bpf_probe_read_str(payload, ...);
if (filepart_length <= MAX_PATH) {
barrier_var(filepart_length); // workaround
payload += filepart_length;
}
}
bpf_iter test failures with clang/llvm 10.0.0
=============================================
With clang/llvm 10.0.0, the following two bpf_iter tests failed:
* ``bpf_iter/ipv6_route``
* ``bpf_iter/netlink``
The symptom for ``bpf_iter/ipv6_route`` looks like
.. code-block:: c
2: (79) r8 = *(u64 *)(r1 +8)
...
14: (bf) r2 = r8
15: (0f) r2 += r1
; BPF_SEQ_PRINTF(seq, "%pi6 %02x ", &rt->fib6_dst.addr, rt->fib6_dst.plen);
16: (7b) *(u64 *)(r8 +64) = r2
only read is supported
The symptom for ``bpf_iter/netlink`` looks like
.. code-block:: c
; struct netlink_sock *nlk = ctx->sk;
2: (79) r7 = *(u64 *)(r1 +8)
...
15: (bf) r2 = r7
16: (0f) r2 += r1
; BPF_SEQ_PRINTF(seq, "%pK %-3d ", s, s->sk_protocol);
17: (7b) *(u64 *)(r7 +0) = r2
only read is supported
This is due to a llvm BPF backend bug. `The fix`__
has been pushed to llvm 10.x release branch and will be
available in 10.0.1. The patch is available in llvm 11.0.0 trunk.
__ https://reviews.llvm.org/D78466
BPF CO-RE-based tests and Clang version
=======================================
A set of selftests use BPF target-specific built-ins, which might require
bleeding-edge Clang versions (Clang 12 nightly at this time).
Few sub-tests of core_reloc test suit (part of test_progs test runner) require
the following built-ins, listed with corresponding Clang diffs introducing
them to Clang/LLVM. These sub-tests are going to be skipped if Clang is too
old to support them, they shouldn't cause build failures or runtime test
failures:
- __builtin_btf_type_id() [0_, 1_, 2_];
- __builtin_preserve_type_info(), __builtin_preserve_enum_value() [3_, 4_].
.. _0: https://reviews.llvm.org/D74572
.. _1: https://reviews.llvm.org/D74668
.. _2: https://reviews.llvm.org/D85174
.. _3: https://reviews.llvm.org/D83878
.. _4: https://reviews.llvm.org/D83242
|
