[Bug c/107348] New: documentation: __builtin_classify_type() undocumented

[Bug c/107348] New: documentation: __builtin_classify_type() undocumented

[colomar.6.4.3 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107348

            Bug ID: 107348
           Summary: documentation: __builtin_classify_type() undocumented
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: colomar.6.4.3 at gmail dot com
  Target Milestone: ---

I can't find any documentation about `__builtin_classify_type()`.

Moreover, it behaves weird with arrays.

$ grepc type_class
./gcc/typeclass.h:30:
enum type_class
{
  no_type_class = -1,
  void_type_class, integer_type_class, char_type_class,
  enumeral_type_class, boolean_type_class,
  pointer_type_class, reference_type_class, offset_type_class,
  real_type_class, complex_type_class,
  function_type_class, method_type_class,
  record_type_class, union_type_class,
  array_type_class, string_type_class,
  lang_type_class, opaque_type_class
};


One would expect that for arrays, it would return `array_type_class`, i.e.,
`14`, but it returns `5`, i.e., `pointer_type_class`.

alx@debian:~/tmp$ cat ptr.c 
#include <stdio.h>

int
main(void)
{
        int a[5];

        printf("%d\n", __builtin_classify_type(a));
}
alx@debian:~/tmp$ cc -Wall -Wextra ptr.c 
alx@debian:~/tmp$ ./a.out 
5

Please document the builtin (and also fix the handling of arrays).

[Bug c/107348] documentation: __builtin_classify_type() undocumented

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107348

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
There is a array to pointer decay in the argument, so pointer_type_class is the
right answer.
As for documentation, seems __builtin_classify_type is documented in gccint
manual rather than gcc:
https://gcc.gnu.org/onlinedocs/gccint/Varargs.html#Varargs

[Bug c/107348] documentation: __builtin_classify_type() undocumented

[colomar.6.4.3 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107348

--- Comment #2 from Alejandro Colomar <colomar.6.4.3 at gmail dot com> ---
Being a builtin, I expected that you could just do "compiler magic" to avoid
the decay.

[Bug c/107348] documentation: __builtin_classify_type() undocumented

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107348

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Most of the builtins don't do such special magic.
It is true that some, mostly newer added, builtins do have such magic, but I
think __builtin_classify_type is one of the first builtins added to GCC
(certainly GCC 1.36 in 1989 had it).
It would be very bad idea to change its behavior all of sudden after 33 years.

[Bug c/107348] documentation: __builtin_classify_type() undocumented

[colomar.6.4.3 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107348

--- Comment #4 from Alejandro Colomar <colomar.6.4.3 at gmail dot com> ---
Hmm.  Then I wonder if array_type_class is being used at all, or if it's unused
code.

[Bug c/107348] documentation: __builtin_classify_type() undocumented

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107348

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Most likely.  I think char_type_class, function_type_class, method_type_class,
string_type_class most likely too (function and methods also decay to pointers
to pointer to members), string has the same issue as array, char isn't even
handled in type_to_class).  Not really sure about reference_type_class, maybe
even that.

[Bug c/107348] documentation: __builtin_classify_type() undocumented

[colomar.6.4.3 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107348

--- Comment #6 from Alejandro Colomar <colomar.6.4.3 at gmail dot com> ---
Thanks!

Maybe a working __builtin_classify_type2() would be useful...  Maybe not...

BTW, maybe it's worth checking all the code in gcc that is comparing the result
against 14, and see if it should be removed, or fixed, or if it actually works
somehow.

Cheers,
Alex

P.S.:  Maybe having a search function, or a single-HTML-page manual (as glibc
does) would be good for finding this stuff.  I don't know.  I tried searching
in the PDF, and I didn't find it.

[Bug c/107348] documentation: __builtin_classify_type() undocumented

[alx at kernel dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107348

Alejandro Colomar <alx at kernel dot org> changed:

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

--- Comment #7 from Alejandro Colomar <alx at kernel dot org> ---
I just realized it was fixed recently.  Thanks!


commit 53d834a7fae3afffebb45a2d66908f705773a7fc
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Sep 20 18:37:29 2023 +0200

    c, c++: Accept __builtin_classify_type (typename)

    As mentioned in my stdckdint.h mail, __builtin_classify_type has
    a problem that argument promotion (the argument is passed to ...
    prototyped builtin function) means that certain type classes will
    simply never appear.
    I think it is too late to change how it behaves, lots of code in the
    wild might rely on the current behavior.

    So, the following patch adds option to use a typename rather than
    expression as the operand to the builtin, making it behave similarly
    to sizeof, typeof or say the clang _Generic extension where the
    first argument can be there not just expression, but also typename.

    I think we have other prior art here, e.g. __builtin_va_arg also
    expects typename.

    I've added this to both C and C++, because it would be weird if it
    supported it only in C and not in C++.

    2023-09-20  Jakub Jelinek  <jakub@redhat.com>

    gcc/
            * builtins.h (type_to_class): Declare.
            * builtins.cc (type_to_class): No longer static.  Return
            int rather than enum.
            * doc/extend.texi (__builtin_classify_type): Document.
    gcc/c/
            * c-parser.cc (c_parser_postfix_expression_after_primary): Parse
            __builtin_classify_type call with typename as argument.
    gcc/cp/
            * parser.cc (cp_parser_postfix_expression): Parse
            __builtin_classify_type call with typename as argument.
            * pt.cc (tsubst_copy_and_build): Handle __builtin_classify_type
            with dependent typename as argument.
    gcc/testsuite/
            * c-c++-common/builtin-classify-type-1.c: New test.
            * g++.dg/ext/builtin-classify-type-1.C: New test.
            * g++.dg/ext/builtin-classify-type-2.C: New test.
            * gcc.dg/builtin-classify-type-1.c: New test.