[Bug c/101125] New: warn when a construct would become invalid if a function were replaced by a function-like macro

[Bug c/101125] New: warn when a construct would become invalid if a function were replaced by a function-like macro

[vincent-gcc at vinc17 dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101125

            Bug ID: 101125
           Summary: warn when a construct would become invalid if a
                    function were replaced by a function-like macro
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: vincent-gcc at vinc17 dot net
  Target Milestone: ---

In the standard C library and some other libraries (like GNU MPFR), functions
may be replaced by function-like macros with the same behavior. So, in order to
detect potential portability issues, GCC should provide a warning to check
whether a construct used with a function call would become invalid if the
function were replaced by a macro.

Such a difference can at least come from the use of the comma in some
expressions. Example:

struct s { int a, b; };
void f (int);
#define F(X) f(X)
void g (void)
{
  f ((struct s){0,1}.a);
  F ((struct s){0,1}.a);
}

The call to f is valid, but the use of the macro yields an error because the
preprocessor assumes that there are 2 arguments in the call: "(struct s){0" and
"1}.a". Here, one should use additional parentheses:

  F (((struct s){0,1}.a));

AFAIK, the library (which provides the function and the macro in a header file)
cannot do anything to avoid such an error; only the user can avoid it, but he
may not be aware of the issue as long as only a function is provided. Hence the
usefulness of the proposed warning.

[Bug c/101125] warn when a construct would become invalid if a function were replaced by a function-like macro

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101125

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
           Keywords|                            |diagnostic

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I really doubt this is an useful warning really.  GCC has already had a notion
that inline functions should be as fast as macro functions and type safe ...

[Bug c/101125] warn when a construct would become invalid if a function were replaced by a function-like macro

[vincent-gcc at vinc17 dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101125

--- Comment #2 from Vincent Lefèvre <vincent-gcc at vinc17 dot net> ---
Even though GCC support fast inline functions, some libraries may still use
function-like macros. In particular, this is still the case with glibc (e.g.
pthread_cleanup_pop).