public inbox for llvm@lists.linux.dev
 help / color / mirror / Atom feed
From: Yonghong Song <yhs@meta.com>
To: James Hilliard <james.hilliard1@gmail.com>,
	Andrii Nakryiko <andrii.nakryiko@gmail.com>
Cc: bpf@vger.kernel.org, Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Andrii Nakryiko <andrii@kernel.org>,
	Martin KaFai Lau <martin.lau@linux.dev>,
	Song Liu <song@kernel.org>, Yonghong Song <yhs@fb.com>,
	John Fastabend <john.fastabend@gmail.com>,
	KP Singh <kpsingh@kernel.org>,
	Stanislav Fomichev <sdf@google.com>, Hao Luo <haoluo@google.com>,
	Jiri Olsa <jolsa@kernel.org>, Mykola Lysenko <mykolal@fb.com>,
	Shuah Khan <shuah@kernel.org>,
	Nathan Chancellor <nathan@kernel.org>,
	Nick Desaulniers <ndesaulniers@google.com>,
	Tom Rix <trix@redhat.com>, Eduard Zingerman <eddyz87@gmail.com>,
	linux-kselftest@vger.kernel.org, linux-kernel@vger.kernel.org,
	llvm@lists.linux.dev
Subject: Re: [PATCH bpf-next] selftests/bpf: move struct definitions out of function params
Date: Wed, 28 Dec 2022 10:42:37 -0800	[thread overview]
Message-ID: <6ba9832d-d7a6-9bfa-324c-c8cbc57fe8aa@meta.com> (raw)
In-Reply-To: <CADvTj4o4RvmjQrO4tjfUf0=Pw7teWmsR_toNSJjWivhD5ukK6g@mail.gmail.com>



On 12/22/22 11:26 AM, James Hilliard wrote:
> On Thu, Dec 22, 2022 at 12:04 PM Andrii Nakryiko
> <andrii.nakryiko@gmail.com> wrote:
>>
>> On Tue, Dec 20, 2022 at 9:59 PM James Hilliard
>> <james.hilliard1@gmail.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@gmail.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
>>>

      reply	other threads:[~2022-12-28 18:43 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-12-21  5:58 [PATCH bpf-next] selftests/bpf: move struct definitions out of function params James Hilliard
2022-12-21 19:07 ` sdf
2022-12-22 19:03 ` Andrii Nakryiko
2022-12-22 19:26   ` James Hilliard
2022-12-28 18:42     ` Yonghong Song [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=6ba9832d-d7a6-9bfa-324c-c8cbc57fe8aa@meta.com \
    --to=yhs@meta.com \
    --cc=andrii.nakryiko@gmail.com \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=eddyz87@gmail.com \
    --cc=haoluo@google.com \
    --cc=james.hilliard1@gmail.com \
    --cc=john.fastabend@gmail.com \
    --cc=jolsa@kernel.org \
    --cc=kpsingh@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-kselftest@vger.kernel.org \
    --cc=llvm@lists.linux.dev \
    --cc=martin.lau@linux.dev \
    --cc=mykolal@fb.com \
    --cc=nathan@kernel.org \
    --cc=ndesaulniers@google.com \
    --cc=sdf@google.com \
    --cc=shuah@kernel.org \
    --cc=song@kernel.org \
    --cc=trix@redhat.com \
    --cc=yhs@fb.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox