[Bug c++/109775] New: gcc misidentifies a VLA

[Bug c++/109775] New: gcc misidentifies a VLA

[janezz55 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109775

            Bug ID: 109775
           Summary: gcc misidentifies a VLA
           Product: gcc
           Version: 13.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: janezz55 at gmail dot com
  Target Milestone: ---

If I compile the example .cpp file:

gcc -std=c++20 -Wextra -Wall -Wpedantic add_sub_tests.cpp -o a

gcc produces the warning:

../intt.hpp: In member function 'constexpr auto intt::intt<
<template-parameter-1-1>, N, <anonymous> >::naimul(const intt::intt<
<template-parameter-1-1>, N, <anonymous> >&) const':
../intt.hpp:675:24: warning: ISO C++ forbids variable length array 'v_' [-Wvla]
  675 |           D const pp(D(v_[i]) * o.v_[S - i]);
      |                        ^~

But no VLA is involved and clang++ 15 manages to avoid this issue.

The example file is here:

https://github.com/user1095108/intt/blob/master/tests/add_sub_tests.cpp

I believe the bug was present in prior versions of gcc as well.

[Bug c++/109775] gcc misidentifies a VLA

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109775

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
It has to do with parsing here:
GCC starts parsing it as a function declaration and then backs out but had
already warned/errored out about the VLA.

If you change the code to:
D const pp{D(v_[i]) * o.v_[S - i]};

Then there is no ambiguity in parsing this.

[Bug c++/109775] gcc misidentifies a VLA

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109775

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Created attachment 55022
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=55022&action=edit
Non reduced testcase

[Bug c++/109775] gcc misidentifies a VLA

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109775

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2023-05-08
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Reduced testcase:
```
template <typename T>
constexpr void naimul(const T v_[2])
{
  int i = 1;
  T pp(T(v_[i])+0);
}
```

[Bug c++/109775] gcc misidentifies a VLA

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109775

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Here is another related testcase:
```
typedef int T;
void naimul(const T v_[2])
{
  int i = 0;
  T pp(T(v_[0])+0);
}
```
We get an -pedantic-error dealing with 0 sized array:
<source>: In function 'void naimul(const T*)':
<source>:5:13: error: ISO C++ forbids zero-size array 'v_' [-Wpedantic]
    5 |   T pp(T(v_[0])+0);
      |             ^

[Bug c++/109775] gcc misidentifies a VLA

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109775

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Depends on|                            |61259

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #4)
> Here is another related testcase:
> ```
> typedef int T;
> void naimul(const T v_[2])
> {
>   int i = 0;
>   T pp(T(v_[0])+0);
> }
> ```
> We get an -pedantic-error dealing with 0 sized array:
> <source>: In function 'void naimul(const T*)':
> <source>:5:13: error: ISO C++ forbids zero-size array 'v_' [-Wpedantic]
>     5 |   T pp(T(v_[0])+0);
>       |             ^

That is recorded as PR 61259 .


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61259
[Bug 61259] [10/11/12/13/14 Regression] Spurious "ISO C++ forbids zero-size
array" warning with -pedantic