[Bug c/107763] New: -Wreturn-type false-positive with fully-covered switch over enum

[Bug c/107763] New: -Wreturn-type false-positive with fully-covered switch over enum

[lebedev.ri at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107763

            Bug ID: 107763
           Summary: -Wreturn-type false-positive with fully-covered switch
                    over enum
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: lebedev.ri at gmail dot com
  Target Milestone: ---

enum a { b };

int foo(a a) {
    switch(a) {
        case b: return 42;
    }
}


The `a` can only have value `a::b`,
so the function always returns 42, yet gcc complains:
> warning: control reaches end of non-void function [-Wreturn-type]

https://godbolt.org/z/j8ezrr5h3

[Bug c/107763] -Wreturn-type false-positive with fully-covered switch over enum

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107763

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |INVALID

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
>The `a` can only have value `a::b`,

NO, it has a full range of the underlying type. For GCC, see
https://gcc.gnu.org/onlinedocs/gcc-12.2.0/gcc/Structures-unions-enumerations-and-bit-fields-implementation.html#Structures-unions-enumerations-and-bit-fields-implementation

"Normally, the type is unsigned int if there are no negative values in the
enumeration, otherwise int. If -fshort-enums is specified, then if there are
negative values it is the first of signed char, short and int that can
represent all the values, otherwise it is the first of unsigned char, unsigned
short and unsigned int that can represent all the values.

On some targets, -fshort-enums is the default; this is determined by the ABI.

"

[Bug c++/107763] -Wreturn-type false-positive with fully-covered switch over enum

[lebedev.ri at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107763

Roman Lebedev <lebedev.ri at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |c++

--- Comment #2 from Roman Lebedev <lebedev.ri at gmail dot com> ---
(In reply to Andrew Pinski from comment #1)
> >The `a` can only have value `a::b`,
> 
> NO, it has a full range of the underlying type. For GCC, see
> https://gcc.gnu.org/onlinedocs/gcc-12.2.0/gcc/Structures-unions-enumerations-
> and-bit-fields-implementation.html#Structures-unions-enumerations-and-bit-
> fields-implementation
> 
> "Normally, the type is unsigned int if there are no negative values in the
> enumeration, otherwise int. If -fshort-enums is specified, then if there are
> negative values it is the first of signed char, short and int that can
> represent all the values, otherwise it is the first of unsigned char,
> unsigned short and unsigned int that can represent all the values.
> 
> On some targets, -fshort-enums is the default; this is determined by the ABI.
> 
> "

Is this situation different in C++? looks like i set the component wrong.
Is this implementation-defined behavior,
or are you saying that clang is wrong here?

At run time, `a` can not have any other value other than `a::b`:
https://godbolt.org/z/hnc665755
so this really does seem like a bug.

[Bug c++/107763] -Wreturn-type false-positive with fully-covered switch over enum

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107763

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
No, it is not different in C++. See PR 91950 for that.

[Bug c++/107763] -Wreturn-type false-positive with fully-covered switch over enum

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107763

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Roman Lebedev from comment #2)> 
> Is this situation different in C++? looks like i set the component wrong.
> Is this implementation-defined behavior,
> or are you saying that clang is wrong here?

Yes I think clang is wrong. The spec says: "The value is unchanged if the
original value is within the range of the enumeration values"

[Bug c++/107763] -Wreturn-type false-positive with fully-covered switch over enum

[lebedev.ri at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107763

--- Comment #5 from Roman Lebedev <lebedev.ri at gmail dot com> ---
Thank you.
Forwarded to https://github.com/llvm/llvm-project/issues/59085