| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Wed, 28 Dec 2022 10:42:37 -0800
From: Yonghong Song <yhs@...a.com>
To: James Hilliard <james.hilliard1@...il.com>,
Andrii Nakryiko <andrii.nakryiko@...il.com>
Cc: bpf@...r.kernel.org, Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Andrii Nakryiko <andrii@...nel.org>,
Martin KaFai Lau <martin.lau@...ux.dev>,
Song Liu <song@...nel.org>, Yonghong Song <yhs@...com>,
John Fastabend <john.fastabend@...il.com>,
KP Singh <kpsingh@...nel.org>,
Stanislav Fomichev <sdf@...gle.com>,
Hao Luo <haoluo@...gle.com>, Jiri Olsa <jolsa@...nel.org>,
Mykola Lysenko <mykolal@...com>, Shuah Khan <shuah@...nel.org>,
Nathan Chancellor <nathan@...nel.org>,
Nick Desaulniers <ndesaulniers@...gle.com>,
Tom Rix <trix@...hat.com>,
Eduard Zingerman <eddyz87@...il.com>,
linux-kselftest@...r.kernel.org, linux-kernel@...r.kernel.org,
llvm@...ts.linux.dev
Subject: Re: [PATCH bpf-next] selftests/bpf: move struct definitions out of
function params
On 12/22/22 11:26 AM, James Hilliard wrote:
> On Thu, Dec 22, 2022 at 12:04 PM Andrii Nakryiko
> <andrii.nakryiko@...il.com> wrote:
>>
>> On Tue, Dec 20, 2022 at 9:59 PM James Hilliard
>> <james.hilliard1@...il.com> wrote:
>>>
>>> Anonymous structs can't be declared inside function parameter
>>> definitions in current c standards, however clang doesn't detect this
>>> condition currently while GCC does.
>>>
>>> Details: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108189
>>>
>>> Fixes errors like:
>>> progs/btf_dump_test_case_bitfields.c:85:7: error: anonymous struct declared inside parameter list will not be visible outside of this definition or declaration [-Werror]
>>> 85 | int f(struct {
>>> | ^~~~~~
>>>
>>> Signed-off-by: James Hilliard <james.hilliard1@...il.com>
>>> ---
>>> .../bpf/progs/btf_dump_test_case_bitfields.c | 9 ++++--
>>> .../progs/btf_dump_test_case_namespacing.c | 10 ++++---
>>> .../bpf/progs/btf_dump_test_case_packing.c | 10 ++++---
>>> .../bpf/progs/btf_dump_test_case_padding.c | 10 ++++---
>>> .../bpf/progs/btf_dump_test_case_syntax.c | 30 +++++++++++++------
>>> 5 files changed, 46 insertions(+), 23 deletions(-)
>>>
>>> diff --git a/tools/testing/selftests/bpf/progs/btf_dump_test_case_bitfields.c b/tools/testing/selftests/bpf/progs/btf_dump_test_case_bitfields.c
>>> index e01690618e1e..c75f6bd06a49 100644
>>> --- a/tools/testing/selftests/bpf/progs/btf_dump_test_case_bitfields.c
>>> +++ b/tools/testing/selftests/bpf/progs/btf_dump_test_case_bitfields.c
>>> @@ -82,11 +82,16 @@ struct bitfield_flushed {
>>> long b: 16;
>>> };
>>>
>>> -int f(struct {
>>> +/* ----- START-EXPECTED-OUTPUT ----- */
>>> +struct root_struct {
>>
>> there is no need to make this struct part of expected output, just
>> keep it next to f?
>
> Seems to be required as the diff check fails otherwise.
>
>>
>>
>>> struct bitfields_only_mixed_types _1;
>>> struct bitfield_mixed_with_others _2;
>>> struct bitfield_flushed _3;
>>> -} *_)
>>> +};
>>> +
>>> +/* ------ END-EXPECTED-OUTPUT ------ */
>>> +
>>> +int f(struct root_struct *_)
>>> {
>>> return 0;
>>> }
>>> diff --git a/tools/testing/selftests/bpf/progs/btf_dump_test_case_namespacing.c b/tools/testing/selftests/bpf/progs/btf_dump_test_case_namespacing.c
>>> index 92a4ad428710..d7cf2a8487c9 100644
>>> --- a/tools/testing/selftests/bpf/progs/btf_dump_test_case_namespacing.c
>>> +++ b/tools/testing/selftests/bpf/progs/btf_dump_test_case_namespacing.c
>>> @@ -49,9 +49,7 @@ typedef int Y;
>>>
>>> typedef int Z;
>>>
>>> -/*------ END-EXPECTED-OUTPUT ------ */
>>> -
>>> -int f(struct {
>>> +struct root_struct {
>>> struct S _1;
>>> S _2;
>>> union U _3;
>>> @@ -67,7 +65,11 @@ int f(struct {
>>> X xx;
>>> Y yy;
>>> Z zz;
>>> -} *_)
>>> +};
>>
>> same, that struct is only to preserve all the referenced types, so
>> keep it hidden from the output
>
> I wasn't able to find a way to keep it out of the output.
>
> The other tests with a root_struct seem to always have it in the output:
> https://github.com/torvalds/linux/blob/v6.1/tools/testing/selftests/bpf/progs/btf_dump_test_case_multidim.c#L21-L28
> https://github.com/torvalds/linux/blob/v6.1/tools/testing/selftests/bpf/progs/btf_dump_test_case_ordering.c#L50-L56
> https://github.com/torvalds/linux/blob/v6.1/tools/testing/selftests/bpf/progs/btf_dump_test_case_syntax.c#L222-L237
>
>>
>>> +
>>> +/*------ END-EXPECTED-OUTPUT ------ */
>>> +
>>> +int f(struct root_struct *_)
>>> {
>>> return 0;
>>> }
>>> diff --git a/tools/testing/selftests/bpf/progs/btf_dump_test_case_packing.c b/tools/testing/selftests/bpf/progs/btf_dump_test_case_packing.c
>>> index 7998f27df7dd..e039ceb50c43 100644
>>> --- a/tools/testing/selftests/bpf/progs/btf_dump_test_case_packing.c
>>> +++ b/tools/testing/selftests/bpf/progs/btf_dump_test_case_packing.c
>>> @@ -132,9 +132,7 @@ struct outer_packed_struct {
>>> struct nested_packed_struct b;
>>> } __attribute__((packed));
>>>
>>> -/* ------ END-EXPECTED-OUTPUT ------ */
>>> -
>>> -int f(struct {
>>> +struct root_struct {
>>> struct packed_trailing_space _1;
>>> struct non_packed_trailing_space _2;
>>> struct packed_fields _3;
>>> @@ -147,7 +145,11 @@ int f(struct {
>>> struct usb_host_endpoint _10;
>>> struct outer_nonpacked_struct _11;
>>> struct outer_packed_struct _12;
>>> -} *_)
>>> +};
>>> +
>>> +/* ------ END-EXPECTED-OUTPUT ------ */
>>> +
>>> +int f(struct root_struct *_)
>>> {
>>> return 0;
>>> }
>>> diff --git a/tools/testing/selftests/bpf/progs/btf_dump_test_case_padding.c b/tools/testing/selftests/bpf/progs/btf_dump_test_case_padding.c
>>> index 79276fbe454a..2ca46ad8d66a 100644
>>> --- a/tools/testing/selftests/bpf/progs/btf_dump_test_case_padding.c
>>> +++ b/tools/testing/selftests/bpf/progs/btf_dump_test_case_padding.c
>>> @@ -220,9 +220,7 @@ struct outer_mixed_but_unpacked {
>>> struct nested_packed b2;
>>> };
>>>
>>> -/* ------ END-EXPECTED-OUTPUT ------ */
>>> -
>>> -int f(struct {
>>> +struct root_struct {
>>> struct padded_implicitly _1;
>>> struct padded_explicitly _2;
>>> struct padded_a_lot _3;
>>> @@ -243,7 +241,11 @@ int f(struct {
>>> struct ib_wc _201;
>>> struct acpi_object_method _202;
>>> struct outer_mixed_but_unpacked _203;
>>> -} *_)
>>> +} __attribute__((packed));
>>> +
>>> +/* ------ END-EXPECTED-OUTPUT ------ */
>>> +
>>> +int f(struct root_struct *_)
>>> {
>>> return 0;
>>> }
>>> diff --git a/tools/testing/selftests/bpf/progs/btf_dump_test_case_syntax.c b/tools/testing/selftests/bpf/progs/btf_dump_test_case_syntax.c
>>> index 26fffb02ed10..3e31df7cecc6 100644
>>> --- a/tools/testing/selftests/bpf/progs/btf_dump_test_case_syntax.c
>>> +++ b/tools/testing/selftests/bpf/progs/btf_dump_test_case_syntax.c
>>> @@ -104,24 +104,24 @@ typedef void (*printf_fn_t)(const char *, ...);
>>> * typedef const fn_output_inner_t fn_ptr_arr2_t[5];
>>> */
>>> /* ----- START-EXPECTED-OUTPUT ----- */
>>> -typedef char * const * (*fn_ptr2_t)(struct {
>>> - int a;
>>> -}, int (*)(int));
>>> +struct struct_a;
>>> +
>>> +typedef char * const * (*fn_ptr2_t)(struct struct_a, int (*)(int));
>>> +
>>> +struct struct_c;
>>> +
>>> +struct struct_h;
>>>
>>> typedef struct {
>>> int a;
>>> - void (*b)(int, struct {
>>> - int c;
>>> - }, union {
>>> + void (*b)(int, struct struct_c, union {
>>> char d;
>>> int e[5];
>>> });
>>> } (*fn_complex_t)(union {
>>> void *f;
>>> char g[16];
>>> -}, struct {
>>> - int h;
>>> -});
>>> +}, struct struct_h);
>>
>> these do test some pieces of libbpf's btf_dump logic, so I'm way more
>> reluctant to remove these. If I understand correctly, this syntax will
>> be eventually supported by GCC, so is there any way to keep these
>> examples as is by requiring C23 mode or something? Or just skipping
>> compiling this one if GCC is used?
>
> I'm not sure, I'm having trouble finding a description in the C23 specification,
> I presume if it is in there then GCC will eventually support it.
>
> Maybe just keep the root_struct changes for now and hold off on this until
> it's clarified that this is valid C23 code or not?
>
> At the moment it appears there's a clang bug here as it shouldn't be valid
> C17 code(which AFAIU is clang's default).
Looks like clang supports anonymous struct parameter all the way back to
c89. I tried clang12 and latest clang16.
$ cat t.c
typedef char * const * (*fn_ptr2_t)(struct {
int a;
}, int (*)(int));
fn_ptr2_t h;
$ clang -S -emit-llvm -g -std=c89 t.c
$ clang -S -emit-llvm -g -std=c11 t.c
$ clang -S -emit-llvm -g -std=c17 t.c
]$ clang -S -emit-llvm -g t.c
$ gcc -v -S t.c
...
GNU C17 (GCC) version 8.5.0 20210514 (Red Hat 8.5.0-17)
(x86_64-redhat-linux)
...
t.c:1:37: warning: anonymous struct declared inside parameter list will
not be visible outside of this definition or declaration
typedef char * const * (*fn_ptr2_t)(struct {
^~~~~~
...
I agree that we should understand better about standard requirement
before making the change.
>
>>
>>>
>>> typedef void (* (*signal_t)(int, void (*)(int)))(int);
>>>
>>> @@ -272,6 +272,18 @@ struct root_struct {
>>> struct float_struct _15;
>>> };
>>>
>>> +struct struct_a {
>>> + int a;
>>> +};
>>> +
>>> +struct struct_h {
>>> + int h;
>>> +};
>>> +
>>> +struct struct_c {
>>> + int c;
>>> +};
>>> +
>>> /* ------ END-EXPECTED-OUTPUT ------ */
>>>
>>> int f(struct root_struct *s)
>>> --
>>> 2.34.1
>>>
Powered by blists - more mailing lists