[Bug bootstrap/101377] New: 'exec format error' on x86_64-w64-mingw32

[Bug bootstrap/101377] New: 'exec format error' on x86_64-w64-mingw32

[lh_mouse at 126 dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

            Bug ID: 101377
           Summary: 'exec format error' on x86_64-w64-mingw32
           Product: gcc
           Version: 11.1.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
          Assignee: unassigned at gcc dot gnu.org
          Reporter: lh_mouse at 126 dot com
  Target Milestone: ---

Created attachment 51117
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51117&action=edit
bootstrap log

With binutils 2.36 I get

 
/d/lh_mouse/GitHub/MINGW-packages-dev/mingw-w64-gcc-git/src/gcc/libgomp/configure:
line 4071:
./conftest.exe: cannot execute binary file: Exec format error

Reverting cfc9fdcec8861be0d11ec4493cca518bf5b4c32d makes the build proceed
again.

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1
           Keywords|                            |build
            Summary|'exec format error' on      |[11/12 Regression] 'exec
                   |x86_64-w64-mingw32          |format error' on
                   |                            |x86_64-w64-mingw32
   Target Milestone|---                         |11.2
      Known to fail|                            |11.1.1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
So it's likely the issue that mingw/cygwin do not like executables with DWARF5
debug and g:cfc9fdcec8861be0d11ec4493cca518bf5b4c32d somehow autodetects
support for it in gas which then, despite GCC defaulting to DWARF4 ends up
emitting DWARF5 by means of --gdwarf-5 passed to it?

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

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

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

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Note this should block 11.2 with a RC due in about two weeks.

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2021-07-08
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |WAITING

--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> So it's likely the issue that mingw/cygwin do not like executables with
> DWARF5 debug and g:cfc9fdcec8861be0d11ec4493cca518bf5b4c32d somehow
> autodetects support for it in gas which then, despite GCC defaulting to DWARF4
> ends up emitting DWARF5 by means of --gdwarf-5 passed to it?

Binutils 2.36 should be detected as HAVE_LD_BROKEN_PE_DWARF5 by the configure
script and DWARF 4 forced by:

#ifdef HAVE_LD_BROKEN_PE_DWARF5
  /* If the PE linker has broken DWARF 5 support, make
     DWARF 4 the default.  */
  if (TARGET_PECOFF)
    SET_OPTION_IF_UNSET (opts, opts_set, dwarf_version, 4);
#endif

so --gdwarf-5 should not be passed, but something obviously does not work.

Could you post the *couple* of lines from config.log starting with:

configure:30285: checking broken PE linker dwarf5 support

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[lh_mouse at 126 dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #4 from Liu Hao <lh_mouse at 126 dot com> ---
configure said 'yes' for 'broken dwarf5 support'. I am not quite clear whether
this means whether it is broken or not (but I suspect yes):

```
configure:29798: checking linker PT_GNU_EH_FRAME support
configure:29829: result: no
configure:29832: checking linker CIEv3 in .eh_frame support
configure:29869: result: yes
configure:29872: checking linker position independent executable support
configure:29913: result: yes
configure:29916: checking linker PIE support with copy reloc
configure:29965: result: no
configure:29968: checking linker EH-compatible garbage collection of sections
configure:30048: result: no
configure:30051: checking linker EH garbage collection of sections bug
configure:30095: result: no
configure:30098: checking linker for compressed debug sections
configure:30158: result: 3
configure:30228: checking broken PE linker dwarf5 support
configure:30249: result: yes
configure:30258: checking linker --as-needed support
configure:30319: result: yes
```

This macro was also defined:

```
| #define HAVE_LD_BROKEN_PE_DWARF5 1
```

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #5 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> configure said 'yes' for 'broken dwarf5 support'. I am not quite clear
> whether this means whether it is broken or not (but I suspect yes):

Yes, it's indeed broken as expected.  Out of curiosity, can you attach this
config.log file as well?

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[lh_mouse at 126 dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #6 from Liu Hao <lh_mouse at 126 dot com> ---
Created attachment 51118
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51118&action=edit
gzip'd gcc/config.log

Of course.

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #7 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> Of course.

Thanks.  There are weird things in the log, for example:

configure:29292: checking assembler for assembly of compiler generated 64-bit
.debug_line
configure:29306: /mingw64/x86_64-w64-mingw32/bin/as    -o conftest.o conftest.s
>&5
conftest.s: Assembler messages:
conftest.s:17: Error: unknown pseudo-op: `.4byte'
conftest.s:18: Error: unknown pseudo-op: `.2byte'
conftest.s:19: Error: unknown pseudo-op: `.4byte'
conftest.s:26: Error: unknown pseudo-op: `.8byte'

.2byte/.4byte/.8byte are supposed to be supported by binutils 2.36.

Can you post the output of /mingw64/x86_64-w64-mingw32/bin/as --version?

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[lh_mouse at 126 dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #8 from LIU Hao <lh_mouse at 126 dot com> ---
lh_mouse@lhmouse-pc ~ $ /mingw64/x86_64-w64-mingw32/bin/as --version
GNU assembler (GNU Binutils) 2.36.1
Copyright (C) 2021 Free Software Foundation, Inc.
This program is free software; you may redistribute it under the terms of
the GNU General Public License version 3 or later.
This program has absolutely no warranty.
This assembler was configured for a target of `x86_64-w64-mingw32'.

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW

--- Comment #9 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> lh_mouse@lhmouse-pc ~ $ /mingw64/x86_64-w64-mingw32/bin/as --version
> GNU assembler (GNU Binutils) 2.36.1
> Copyright (C) 2021 Free Software Foundation, Inc.
> This program is free software; you may redistribute it under the terms of
> the GNU General Public License version 3 or later.
> This program has absolutely no warranty.
> This assembler was configured for a target of `x86_64-w64-mingw32'.

Thanks.  So I was confused and 2.36 indeed rejects the directive on Windows,
you need the upcoming 2.37 to accept them.

In any case, I can reproduce the problem with a cross-compiler.

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |ebotcazou at gcc dot gnu.org
             Status|NEW                         |ASSIGNED

--- Comment #10 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Fixing.

[Bug bootstrap/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #11 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Created attachment 51119
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51119&action=edit
Tentative fix

This is a minimal fix to restore the previous state on Windows (I don't think
that a more complete fix is worth the hassle since binutils 2.37 supports DWARF
5).

[Bug target/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[lh_mouse at 126 dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #12 from LIU Hao <lh_mouse at 126 dot com> ---
I will try rebuilding tomorrow (~11hrs later).

[Bug target/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #13 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> I will try rebuilding tomorrow (~11hrs later).

OK, thanks in advance.

[Bug target/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mckelvey at maskull dot com

--- Comment #14 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 101369 has been marked as a duplicate of this bug. ***

[Bug target/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[lh_mouse at 126 dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #15 from LIU Hao <lh_mouse at 126 dot com> ---
I have bootstrapped GCC 11.1.1 on both {x86_64,i686}-w64-mingw32 and seen no
problem so far.

[Bug target/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #16 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Eric Botcazou <ebotcazou@gcc.gnu.org>:

https://gcc.gnu.org/g:e37e175e7dc3376244eb7705269ada6755616831

commit r12-2189-ge37e175e7dc3376244eb7705269ada6755616831
Author: Eric Botcazou <ebotcazou@adacore.com>
Date:   Fri Jul 9 12:08:52 2021 +0200

    Fix build failure on Windows with older binutils

    This is the build failure on Windows with binutils for which GNU as accepts
    the --gdwarf-5 switch but GNU ld generates broken binaries with DWARF 5.

    We already have the HAVE_LD_BROKEN_PE_DWARF5 kludge to disable DWARF 5 in
    this case but it only tames the DWARF version in the compiler, so the
    driver still passes --gdwarf-5 when invoked on an assembly file with -g.

    gcc/
            PR target/101377
            * gcc.c (ASM_DEBUG_DWARF_OPTION): Set again to --gdwarf2 in
            the case where HAVE_AS_WORKING_DWARF_N_FLAG is not defined
            and HAVE_LD_BROKEN_PE_DWARF5 is defined.

[Bug target/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

--- Comment #17 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Eric Botcazou
<ebotcazou@gcc.gnu.org>:

https://gcc.gnu.org/g:d9d954ebd507572a460f93f404bba71e271e2dda

commit r11-8711-gd9d954ebd507572a460f93f404bba71e271e2dda
Author: Eric Botcazou <ebotcazou@adacore.com>
Date:   Fri Jul 9 12:08:52 2021 +0200

    Fix build failure on Windows with older binutils

    This is the build failure on Windows with binutils for which GNU as accepts
    the --gdwarf-5 switch but GNU ld generates broken binaries with DWARF 5.

    We already have the HAVE_LD_BROKEN_PE_DWARF5 kludge to disable DWARF 5 in
    this case but it only tames the DWARF version in the compiler, so the
    driver still passes --gdwarf-5 when invoked on an assembly file with -g.

    gcc/
            PR target/101377
            * gcc.c (ASM_DEBUG_DWARF_OPTION): Set again to --gdwarf2 in
            the case where HAVE_AS_WORKING_DWARF_N_FLAG is not defined
            and HAVE_LD_BROKEN_PE_DWARF5 is defined.

[Bug target/101377] [11/12 Regression] 'exec format error' on x86_64-w64-mingw32

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101377

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

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

--- Comment #18 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Thanks for reporting the problem.