GNU bug report logs - #35410
27.0.50; 32-bit build of master failing with MSYS2

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 35410 in the body.
You can then email your comments to 35410 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: martin rudalics <rudalics <at> gmx.at>
To: Bug-Gnu-Emacs <bug-gnu-emacs <at> gnu.org>
Subject: 27.0.50; 32-bit build of master failing with MSYS2
Date: Wed, 24 Apr 2019 09:28:50 +0200

Trying a 32-bit MSYS2 build of master on Windows 10 configured with
CFLAGS='-O3' gets me here:

w32.o:w32.c:(.text+0xa29f): undefined reference to `_imp___futime32'
collect2.exe: error: ld returned 1 exit status
make[1]: *** [Makefile:643: temacs.exe] Fehler 1
make[1]: Verzeichnis „/c/emacs/trunk/opt-32/src“ wird verlassen
make: *** [Makefile:423: src] Fehler 2

A build configured with CFLAGS='-O0 -g3' proceeds without problems.
The 32-builds use gcc 7.4.0.  64-bit builds with the same options
proceed without problems using gcc 8.3.0.

A workaround here is to add a check to mingw_time.h like

/* The @#$%^&! MinGW developers stopped updating the values of
   __MINGW32_VERSION, __MINGW32_MAJOR_VERSION, and
   __MINGW32_MINOR_VERSION values in v4.x of the runtime, to
   "discourage its uses".  So the values of those macros can no longer
   be trusted, and we need the workaround below, to have a single set
   of macros we can trust.  (The .17 minor version is arbitrary.)  */
#ifdef __MINGW32__
#include <_mingw.h>
# ifndef __CRT__NO_INLINE
#  define __CRT__NO_INLINE
# endif
#endif

martin

Message #8 received at 35410 <at> debbugs.gnu.org (full text, mbox):

From: Noam Postavsky <npostavs <at> gmail.com>
To: martin rudalics <rudalics <at> gmx.at>
Cc: 35410 <at> debbugs.gnu.org
Subject: Re: bug#35410: 27.0.50; 32-bit build of master failing with MSYS2
Date: Thu, 11 Jul 2019 14:23:42 -0400

martin rudalics <rudalics <at> gmx.at> writes:

> Trying a 32-bit MSYS2 build of master on Windows 10 configured with
> CFLAGS='-O3' gets me here:
>
> w32.o:w32.c:(.text+0xa29f): undefined reference to `_imp___futime32'
> collect2.exe: error: ld returned 1 exit status
> make[1]: *** [Makefile:643: temacs.exe] Fehler 1
> make[1]: Verzeichnis „/c/emacs/trunk/opt-32/src“ wird verlassen
> make: *** [Makefile:423: src] Fehler 2
>
> A build configured with CFLAGS='-O0 -g3' proceeds without problems.
> The 32-builds use gcc 7.4.0.  64-bit builds with the same options
> proceed without problems using gcc 8.3.0.

Seems to work okay here, but my ming32 gcc is 7.2.0.  Is this still a
problem?

Message #11 received at 35410 <at> debbugs.gnu.org (full text, mbox):

From: martin rudalics <rudalics <at> gmx.at>
To: Noam Postavsky <npostavs <at> gmail.com>
Cc: 35410 <at> debbugs.gnu.org
Subject: Re: bug#35410: 27.0.50; 32-bit build of master failing with MSYS2
Date: Fri, 12 Jul 2019 10:12:01 +0200

>> w32.o:w32.c:(.text+0xa29f): undefined reference to `_imp___futime32'
>> collect2.exe: error: ld returned 1 exit status
>> make[1]: *** [Makefile:643: temacs.exe] Fehler 1
>> make[1]: Verzeichnis „/c/emacs/trunk/opt-32/src“ wird verlassen
>> make: *** [Makefile:423: src] Fehler 2
>>
>> A build configured with CFLAGS='-O0 -g3' proceeds without problems.
>> The 32-builds use gcc 7.4.0.  64-bit builds with the same options
>> proceed without problems using gcc 8.3.0.
>
> Seems to work okay here, but my ming32 gcc is 7.2.0.  Is this still a
> problem?

Yes.  With slight changes in the Makefile references:

  CCLD     temacs.exe
w32.o:w32.c:(.text+0xa27f): undefined reference to `_imp___futime32'
collect2.exe: error: ld returned 1 exit status
make[1]: *** [Makefile:651: temacs.exe] Fehler 1
make[1]: Verzeichnis „/c/emacs/trunk/opt-32/src“ wird verlassen
make: *** [Makefile:424: src] Fehler 2

In an earlier conversation with Eli I wrote

> > An optimized 32-bit build apparently does not recognize
> >
> > #ifndef __MINGW64_VERSION_MAJOR

and he told me

> So I guess we need to find an alternative to
> __MINGW64_VERSION_MAJOR that will work in the 32-bit MinGW64 builds.
> Look for macros that include "_VERSION_" or "_MAJOR" in the MinGW64
> headers.

but I had no good idea how to continue.

martin

Message #14 received at 35410 <at> debbugs.gnu.org (full text, mbox):

From: Noam Postavsky <npostavs <at> gmail.com>
To: martin rudalics <rudalics <at> gmx.at>
Cc: 35410 <at> debbugs.gnu.org, Noam Postavsky <npostavs <at> gmail.com>
Subject: Re: bug#35410: 27.0.50; 32-bit build of master failing with MSYS2
Date: Fri, 12 Jul 2019 11:28:25 -0400

martin rudalics <rudalics <at> gmx.at> writes:

>>> w32.o:w32.c:(.text+0xa29f): undefined reference to `_imp___futime32'
>>> collect2.exe: error: ld returned 1 exit status
>>> make[1]: *** [Makefile:643: temacs.exe] Fehler 1
>>> make[1]: Verzeichnis „/c/emacs/trunk/opt-32/src“ wird verlassen
>>> make: *** [Makefile:423: src] Fehler 2
>>>
>>> A build configured with CFLAGS='-O0 -g3' proceeds without problems.
>>> The 32-builds use gcc 7.4.0.  64-bit builds with the same options
>>> proceed without problems using gcc 8.3.0.
>>
>> Seems to work okay here, but my ming32 gcc is 7.2.0.

Hmm, I tried updating, but now my mingw32 gcc is 9.1.0, and it still
builds fine.  I'm not sure if it's practical to support old compiler
versions from a rolling release.

Message #17 received at 35410 <at> debbugs.gnu.org (full text, mbox):

From: martin rudalics <rudalics <at> gmx.at>
To: Noam Postavsky <npostavs <at> gmail.com>
Cc: 35410 <at> debbugs.gnu.org
Subject: Re: bug#35410: 27.0.50; 32-bit build of master failing with MSYS2
Date: Fri, 12 Jul 2019 18:52:32 +0200

> Hmm, I tried updating, but now my mingw32 gcc is 9.1.0, and it still
> builds fine.  I'm not sure if it's practical to support old compiler
> versions from a rolling release.

The problem is not with the compiler version per se but with the
headers shipped with MinGW64.  So hopefully they fixed that problem in
our sense in the meantime.  I'm reluctant to update as well because
the 64-bit builds work sufficiently well and updating always bears the
risk spoiling that.  I filed the report only to hear opinions from
other MinGW64 users but apparently neither of them uses my version.

martin

Message #22 received at 35410-done <at> debbugs.gnu.org (full text, mbox):

From: martin rudalics <rudalics <at> gmx.at>
To: Noam Postavsky <npostavs <at> gmail.com>
Cc: 35410-done <at> debbugs.gnu.org
Subject: Re: bug#35410: 27.0.50; 32-bit build of master failing with MSYS2
Date: Wed, 9 Oct 2019 20:11:38 +0200

>  > Hmm, I tried updating, but now my mingw32 gcc is 9.1.0, and it still
>  > builds fine.  I'm not sure if it's practical to support old compiler
>  > versions from a rolling release.
>
> The problem is not with the compiler version per se but with the
> headers shipped with MinGW64.  So hopefully they fixed that problem in
> our sense in the meantime.  I'm reluctant to update as well because
> the 64-bit builds work sufficiently well and updating always bears the
> risk spoiling that.  I filed the report only to hear opinions from
> other MinGW64 users but apparently neither of them uses my version.

I now upgraded MSYS2 to its present version and it builds fine so I'm
hopefully closing this bug.

Thanks for the heads-up, martin

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.