[Bug preprocessor/108244] New: [12/13 Regression] Cannot preprocess standard <algorithm> header with -E -fdirectives-only

[Bug preprocessor/108244] New: [12/13 Regression] Cannot preprocess standard <algorithm> header with -E -fdirectives-only

[gwhitneycom5 at pobox dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

            Bug ID: 108244
           Summary: [12/13 Regression] Cannot preprocess standard
                    <algorithm> header with -E -fdirectives-only
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: preprocessor
          Assignee: unassigned at gcc dot gnu.org
          Reporter: gwhitneycom5 at pobox dot com
  Target Milestone: ---

Create a file `algoproblem.cxx` containing just:

```
#include <algorithm>
int main() { return 0; }
```

Now `g++-13 -E -fdirectives-only algoproblem.cxx` fails with:

```
In file included from /usr/include/c++/13/bits/stl_pair.h:60,
                 from /usr/include/c++/13/bits/stl_algobase.h:64,
                 from /usr/include/c++/13/algorithm:60,
                 from algoproblem.cxx:1:
/usr/include/c++/13/type_traits:2948:25: error: missing binary operator before
token "("
 2948 |                bool _Nothrow = noexcept(_S_conv<_Tp>(_S_get())),
      |
```

whereas `g++-12 -E -fdirectives-only algoproblem.cxx` runs to completion,
producing plausible output.

Context: I am running the build2 build system, which separates partial
preprocessing with `-E -fdirectives-only` from the main compilation. Thus,
build2 has broken when trying to switch to g++-13. I am running on an OpenSUSE
Tumbleweed linux installation, reasonably up-to-date, and `g++-13 --version`
reports:

```
g++-13 (SUSE Linux) 13.0.0 20221213 (experimental) [revision
0a43f7b1a73c8e3b9cefffe430274d0a3d6d3291]
Copyright (C) 2022 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
```

[Bug preprocessor/108244] [12/13 Regression] Cannot preprocess standard <algorithm> header with -E -fdirectives-only

[gwhitneycom5 at pobox dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

--- Comment #1 from Glen Whitney <gwhitneycom5 at pobox dot com> ---
The problem is present in g++-13 but not g++-12; I wasn't sure if that meant I
should write `[12/13 Regression]` or `[13 Regression]` in the title, sorry. If
I got it wrong just let me know and I will be happy to fix.

[Bug preprocessor/108244] [13 Regression] Cannot preprocess standard <algorithm> header with -E -fdirectives-only

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |13.0
           Keywords|                            |ice-on-valid-code
            Summary|[12/13 Regression] Cannot   |[13 Regression] Cannot
                   |preprocess standard         |preprocess standard
                   |<algorithm> header with -E  |<algorithm> header with -E
                   |-fdirectives-only           |-fdirectives-only
      Known to work|                            |12.2.0

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
In file included from
/opt/compiler-explorer/gcc-trunk-20221228/include/c++/13.0.0/bits/stl_pair.h:60,
                 from
/opt/compiler-explorer/gcc-trunk-20221228/include/c++/13.0.0/bits/stl_algobase.h:64,
                 from
/opt/compiler-explorer/gcc-trunk-20221228/include/c++/13.0.0/algorithm:60,
                 from <source>:2:
/opt/compiler-explorer/gcc-trunk-20221228/include/c++/13.0.0/type_traits:2948:25:
error: missing binary operator before token "("
 2948 |                bool _Nothrow = noexcept(_S_conv<_Tp>(_S_get())),
      |                         ^
/opt/compiler-explorer/gcc-trunk-20221228/include/c++/13.0.0/type_traits:3033:27:
internal compiler error: unspellable token PRAGMA_EOL
 3033 |     ~__nonesuch() = delete;
      |                           ^
0xd431ba c_cpp_diagnostic(cpp_reader*, cpp_diagnostic_level,
cpp_warning_reason, rich_location*, char const*, __va_list_tag (*) [1])
        ???:0
0x238e7ba cpp_error(cpp_reader*, cpp_diagnostic_level, char const*, ...)
        ???:0
0x2399043 cpp_spell_token(cpp_reader*, cpp_token const*, unsigned char*, bool)
        ???:0
0x2399e40 cpp_token_as_text(cpp_reader*, cpp_token const*)
        ???:0
0x239333f _cpp_parse_expr

[Bug preprocessor/108244] [13 Regression] Cannot preprocess standard <algorithm> header with -E -fdirectives-only

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Without -fdirectives-only, it works ...

[Bug preprocessor/108244] [13 Regression] Cannot preprocess standard <algorithm> header with -E -fdirectives-only

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |13.0

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
The reduced header case is just:
#include <type_traits>
int main() { return 0; }

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2022-12-28
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
           Keywords|                            |diagnostic,
                   |                            |needs-bisection,
                   |                            |rejects-valid
            Summary|[13 Regression] Cannot      |[13 Regression] `pragma GCC
                   |preprocess standard         |diagnostic` and -E
                   |<algorithm> header with -E  |-fdirectives-only causes
                   |-fdirectives-only           |the preprocessor to become
                   |                            |confused

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Reduced testcase that shows the issue (different warning/error message though):
```
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wctor-dtor-privacy"
#ifdef t
#endif
```
This gives:
<source>:1:9: warning: extra tokens at end of #ifdef directive
    1 | #pragma GCC diagnostic push
      |         ^
Compiler returned: 0

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused

[lhyatt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

Lewis Hyatt <lhyatt at gcc dot gnu.org> changed:

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

--- Comment #6 from Lewis Hyatt <lhyatt at gcc dot gnu.org> ---
I am taking a look.

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused

[lhyatt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

--- Comment #7 from Lewis Hyatt <lhyatt at gcc dot gnu.org> ---
The testcases mentioned so far in this PR (the original report, and the
reduction in Comment 5) started failing in r13-1544, because that revision
added #pragma GCC diagnostic to the set of pragmas that are registered in
preprocess-only mode, including in -fdirectives-only. But the same issue has
existed always it seems for other pragmas similarly registered; for instance
this testcase fails since gcc 9 and before:

=====
$ cat t2.cpp
#pragma message "hello"
#ifdef t
#endif

$ gcc-9 -E -fdirectives-only t2.cpp > /dev/null
t2.cpp:1:9: warning: extra tokens at end of #ifdef directive
    1 | #pragma message "hello"
      |         ^
=====

I think with -fdirectives-only, #pragma is not meant to be registered or
processed.

The below patch fixes all testcases mentioned so far:

=======
diff --git a/gcc/c-family/c-pragma.cc b/gcc/c-family/c-pragma.cc
index 142a46441ac..e9bcd1869d0 100644
--- a/gcc/c-family/c-pragma.cc
+++ b/gcc/c-family/c-pragma.cc
@@ -1647,7 +1647,8 @@ c_register_pragma_1 (const char *space, const char *name,

   if (flag_preprocess_only)
     {
-      if (!(allow_expansion || ihandler.early_handler.handler_1arg))
+      if (cpp_get_options (parse_in)->directives_only
+         || !(allow_expansion || ihandler.early_handler.handler_1arg))
        return;

       pragma_pp_data pp_data;
========

However it doesn't fix an analogous one using "#pragma omp" even though it
suppresses the registration of the pragma there too. I will track that down and
follow up with a complete patch.

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused

[lhyatt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

--- Comment #8 from Lewis Hyatt <lhyatt at gcc dot gnu.org> ---
Here is the patch:
https://gcc.gnu.org/pipermail/gcc-patches/2022-December/609275.html

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused since r13-1544-ge46f4d7430c52104

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|needs-bisection             |
            Summary|[13 Regression] `pragma GCC |[13 Regression] `pragma GCC
                   |diagnostic` and -E          |diagnostic` and -E
                   |-fdirectives-only causes    |-fdirectives-only causes
                   |the preprocessor to become  |the preprocessor to become
                   |confused                    |confused since
                   |                            |r13-1544-ge46f4d7430c52104
                 CC|                            |marxin at gcc dot gnu.org

--- Comment #9 from Martin Liška <marxin at gcc dot gnu.org> ---
Btw. started with r13-1544-ge46f4d7430c52104.

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused since r13-1544-ge46f4d7430c52104

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused since r13-1544-ge46f4d7430c52104

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |thiago at kde dot org

--- Comment #10 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 108372 has been marked as a duplicate of this bug. ***

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused since r13-1544-ge46f4d7430c52104

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

--- Comment #11 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Lewis Hyatt <lhyatt@gcc.gnu.org>:

https://gcc.gnu.org/g:9ca4899144de6db61a782b03a1257489bd26f750

commit r13-5114-g9ca4899144de6db61a782b03a1257489bd26f750
Author: Lewis Hyatt <lhyatt@gmail.com>
Date:   Thu Dec 29 16:55:21 2022 -0500

    preprocessor: Don't register pragmas in directives-only mode [PR108244]

    libcpp's directives-only mode does not expect deferred pragmas to be
    registered, but to date the c-family registration process has not checked
for
    this case. That issue became more visible since r13-1544, which added the
    commonly used GCC diagnostic pragmas to the set of those registered in
    preprocessing modes. Fix it by checking for directives-only mode in
    c-family/c-pragma.cc.

    gcc/c-family/ChangeLog:

            PR preprocessor/108244
            * c-pragma.cc (c_register_pragma_1): Don't attempt to register any
            deferred pragmas if -fdirectives-only.
            (init_pragma): Likewise.

    gcc/testsuite/ChangeLog:

            * c-c++-common/cpp/pr108244-1.c: New test.
            * c-c++-common/cpp/pr108244-2.c: New test.
            * c-c++-common/gomp/pr108244-3.c: New test.

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused since r13-1544-ge46f4d7430c52104

[lhyatt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

Lewis Hyatt <lhyatt at gcc dot gnu.org> changed:

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

--- Comment #12 from Lewis Hyatt <lhyatt at gcc dot gnu.org> ---
Fixed by r13-5114.

[Bug preprocessor/108244] [13 Regression] `pragma GCC diagnostic` and -E -fdirectives-only causes the preprocessor to become confused since r13-1544-ge46f4d7430c52104

[lhyatt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108244

Lewis Hyatt <lhyatt at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jpewhacker at gmail dot com

--- Comment #13 from Lewis Hyatt <lhyatt at gcc dot gnu.org> ---
*** Bug 91152 has been marked as a duplicate of this bug. ***