bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Ken Brown]

Starting with bzr revision 113315, the build on 64-bit Cygwin fails as 
follows:

gcc -std=gnu99 -Demacs  -I. -I/c/src/emacs/64testt/src  -I../lib 
-I/c/src/emacs/64testt/src/../lib    -D_REENTRANT -I/usr/include/gtk-3.0 
-I/usr/include/atk-1.0 -I/usr/include/at-spi2-atk/2.0 
-I/usr/include/pango-1.0 -I/usr/include/gio-unix-2.0/ 
-I/usr/include/cairo -I/usr/include/gdk-pixbuf-2.0 
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include 
-I/usr/include/harfbuzz -I/usr/include/freetype2 -I/usr/include/pixman-1 
-I/usr/include/libpng15   -DGDK_DISABLE_DEPRECATION_WARNINGS 
-DGLIB_DISABLE_DEPRECATION_WARNINGS -I/usr/include/freetype2 
-D_REENTRANT -I/usr/include/librsvg-2.0 -I/usr/include/glib-2.0 
-I/usr/lib/glib-2.0/include -I/usr/include/gdk-pixbuf-2.0 
-I/usr/include/cairo -I/usr/include/libpng15 -I/usr/include/pixman-1 
-I/usr/include/freetype2   -fopenmp -I/usr/include/ImageMagick 
-I/usr/include/libxml2   -I/usr/include/dbus-1.0 
-I/usr/lib/dbus-1.0/include      -D_REENTRANT -I/usr/include/glib-2.0 
-I/usr/lib/glib-2.0/include   -D_REENTRANT -I/usr/include/gconf/2 
-I/usr/include/dbus-1.0 -I/usr/lib/dbus-1.0/include 
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include 
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include 
-I/usr/include/freetype2       -MMD -MF deps/.d -MP 
-I/usr/include/p11-kit-1   -D_REENTRANT -I/usr/include/glib-2.0 
-I/usr/lib/glib-2.0/include      -O0 -ggdb    \
   -o temacs  vm-limit.o dispnew.o frame.o scroll.o xdisp.o menu.o 
xmenu.o window.o  charset.o coding.o category.o ccl.o character.o 
chartab.o bidi.o  cm.o term.o terminal.o xfaces.o xterm.o xfns.o 
xselect.o xrdb.o xsmfns.o xsettings.o gtkutil.o emacsgtkfixed.o 
dbusbind.o  emacs.o keyboard.o macros.o keymap.o sysdep.o  buffer.o 
filelock.o insdel.o marker.o  minibuf.o fileio.o dired.o  cmds.o 
casetab.o casefiddle.o indent.o search.o regex.o undo.o  alloc.o data.o 
doc.o editfns.o callint.o  eval.o floatfns.o fns.o font.o print.o 
lread.o  syntax.o unexcw.o bytecode.o  process.o gnutls.o callproc.o 
region-cache.o sound.o atimer.o  doprnt.o intervals.o textprop.o 
composite.o xml.o gfilenotify.o  profiler.o     sheap.o cygw32.o xfont.o 
ftfont.o xftfont.o ftxfont.o   fontset.o fringe.o image.o xgselect.o 
terminfo.o  gmalloc.o   lastfile.o   ../lib/libgnu.a       -ltiff -ljpeg 
-lpng -lz -lm -lgif -lXpm  -lgtk-3 -latk-bridge-2.0 -lgdk-3 -latk-1.0 
-lXinerama -lXi -lXrandr -lXcursor -lXcomposite -lXdamage 
-lpangocairo-1.0 -lgdk_pixbuf-2.0 -lcairo-gobject -lgio-2.0 -lXfixes 
-lpangoft2-1.0 -lharfbuzz -lpango-1.0 -lcairo -lpixman-1 -lfontconfig 
-lexpat -lfreetype -lbz2 -lxcb-shm -lxcb-render -lXrender -lXext -lX11 
-lxcb -lXau -lXdmcp -lpng15 -lm -lz -lgmodule-2.0 -lgobject-2.0 -lffi 
-lglib-2.0 -lintl -liconv -lpcre   -lSM -lICE -lX11 -lXrender -lXft 
-lXrender -lfontconfig -lexpat -lfreetype -lz -lbz2 -lX11 -lxcb -lXau 
-lXdmcp     -lrsvg-2 -lgio-2.0 -lgdk_pixbuf-2.0 -lcairo -lgmodule-2.0 
-lgobject-2.0 -lffi -lpixman-1 -lfontconfig -lexpat -lfreetype -lbz2 
-lpng15 -lm -lz -lxcb-shm -lxcb-render -lXrender -lXext -lglib-2.0 
-lintl -liconv -lpcre -lX11 -lxcb -lXau -lXdmcp   -lMagickWand 
-lMagickCore         -ldbus-1 -lpthread -lrt     -lXrandr -lXext 
-lXrender -lX11 -lxcb -lXau -lXdmcp   -lXinerama -lXext -lX11 -lxcb 
-lXau -lXdmcp    -lxml2 -lz -liconv -lm       -lncurses  -lgio-2.0 -lz 
-lgmodule-2.0 -lgobject-2.0 -lffi -lglib-2.0 -lintl -liconv -lpcre 
-lgconf-2 -ldbus-1 -lpthread -lrt -lgio-2.0 -lz -lgmodule-2.0 
-lgobject-2.0 -lffi -lglib-2.0 -lintl -liconv -lpcre   -lgobject-2.0 
-lffi -lglib-2.0 -lintl -liconv -lpcre     -lfreetype -lz -lbz2 
-lfontconfig -lexpat -lfreetype -lz -lbz2      -lgnutls -lnettle 
-lhogweed -lgmp -lintl -liconv -ltasn1 -lp11-kit -lz   -lpthread 
-lgio-2.0 -lz -lgmodule-2.0 -lgobject-2.0 -lffi -lglib-2.0 -lintl 
-liconv -lpcre
../lib/libgnu.a(pipe2.o): In function `rpl_pipe2':
/c/src/emacs/64testt/lib/pipe2.c:144: undefined reference to `setmode'


I think the issue here is that pipe2.c shouldn't be compiled on Cygwin.

Ken

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Ken Brown]

On 7/8/2013 11:47 AM, Ken Brown wrote:
> Starting with bzr revision 113315, the build on 64-bit Cygwin fails as
> follows:
>
> gcc -std=gnu99 -Demacs  -I. -I/c/src/emacs/64testt/src  -I../lib
> -I/c/src/emacs/64testt/src/../lib    -D_REENTRANT -I/usr/include/gtk-3.0
> -I/usr/include/atk-1.0 -I/usr/include/at-spi2-atk/2.0
> -I/usr/include/pango-1.0 -I/usr/include/gio-unix-2.0/
> -I/usr/include/cairo -I/usr/include/gdk-pixbuf-2.0
> -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include
> -I/usr/include/harfbuzz -I/usr/include/freetype2 -I/usr/include/pixman-1
> -I/usr/include/libpng15   -DGDK_DISABLE_DEPRECATION_WARNINGS
> -DGLIB_DISABLE_DEPRECATION_WARNINGS -I/usr/include/freetype2
> -D_REENTRANT -I/usr/include/librsvg-2.0 -I/usr/include/glib-2.0
> -I/usr/lib/glib-2.0/include -I/usr/include/gdk-pixbuf-2.0
> -I/usr/include/cairo -I/usr/include/libpng15 -I/usr/include/pixman-1
> -I/usr/include/freetype2   -fopenmp -I/usr/include/ImageMagick
> -I/usr/include/libxml2   -I/usr/include/dbus-1.0
> -I/usr/lib/dbus-1.0/include      -D_REENTRANT -I/usr/include/glib-2.0
> -I/usr/lib/glib-2.0/include   -D_REENTRANT -I/usr/include/gconf/2
> -I/usr/include/dbus-1.0 -I/usr/lib/dbus-1.0/include
> -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include
> -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include
> -I/usr/include/freetype2       -MMD -MF deps/.d -MP
> -I/usr/include/p11-kit-1   -D_REENTRANT -I/usr/include/glib-2.0
> -I/usr/lib/glib-2.0/include      -O0 -ggdb    \
>    -o temacs  vm-limit.o dispnew.o frame.o scroll.o xdisp.o menu.o
> xmenu.o window.o  charset.o coding.o category.o ccl.o character.o
> chartab.o bidi.o  cm.o term.o terminal.o xfaces.o xterm.o xfns.o
> xselect.o xrdb.o xsmfns.o xsettings.o gtkutil.o emacsgtkfixed.o
> dbusbind.o  emacs.o keyboard.o macros.o keymap.o sysdep.o  buffer.o
> filelock.o insdel.o marker.o  minibuf.o fileio.o dired.o  cmds.o
> casetab.o casefiddle.o indent.o search.o regex.o undo.o  alloc.o data.o
> doc.o editfns.o callint.o  eval.o floatfns.o fns.o font.o print.o
> lread.o  syntax.o unexcw.o bytecode.o  process.o gnutls.o callproc.o
> region-cache.o sound.o atimer.o  doprnt.o intervals.o textprop.o
> composite.o xml.o gfilenotify.o  profiler.o     sheap.o cygw32.o xfont.o
> ftfont.o xftfont.o ftxfont.o   fontset.o fringe.o image.o xgselect.o
> terminfo.o  gmalloc.o   lastfile.o   ../lib/libgnu.a       -ltiff -ljpeg
> -lpng -lz -lm -lgif -lXpm  -lgtk-3 -latk-bridge-2.0 -lgdk-3 -latk-1.0
> -lXinerama -lXi -lXrandr -lXcursor -lXcomposite -lXdamage
> -lpangocairo-1.0 -lgdk_pixbuf-2.0 -lcairo-gobject -lgio-2.0 -lXfixes
> -lpangoft2-1.0 -lharfbuzz -lpango-1.0 -lcairo -lpixman-1 -lfontconfig
> -lexpat -lfreetype -lbz2 -lxcb-shm -lxcb-render -lXrender -lXext -lX11
> -lxcb -lXau -lXdmcp -lpng15 -lm -lz -lgmodule-2.0 -lgobject-2.0 -lffi
> -lglib-2.0 -lintl -liconv -lpcre   -lSM -lICE -lX11 -lXrender -lXft
> -lXrender -lfontconfig -lexpat -lfreetype -lz -lbz2 -lX11 -lxcb -lXau
> -lXdmcp     -lrsvg-2 -lgio-2.0 -lgdk_pixbuf-2.0 -lcairo -lgmodule-2.0
> -lgobject-2.0 -lffi -lpixman-1 -lfontconfig -lexpat -lfreetype -lbz2
> -lpng15 -lm -lz -lxcb-shm -lxcb-render -lXrender -lXext -lglib-2.0
> -lintl -liconv -lpcre -lX11 -lxcb -lXau -lXdmcp   -lMagickWand
> -lMagickCore         -ldbus-1 -lpthread -lrt     -lXrandr -lXext
> -lXrender -lX11 -lxcb -lXau -lXdmcp   -lXinerama -lXext -lX11 -lxcb
> -lXau -lXdmcp    -lxml2 -lz -liconv -lm       -lncurses  -lgio-2.0 -lz
> -lgmodule-2.0 -lgobject-2.0 -lffi -lglib-2.0 -lintl -liconv -lpcre
> -lgconf-2 -ldbus-1 -lpthread -lrt -lgio-2.0 -lz -lgmodule-2.0
> -lgobject-2.0 -lffi -lglib-2.0 -lintl -liconv -lpcre   -lgobject-2.0
> -lffi -lglib-2.0 -lintl -liconv -lpcre     -lfreetype -lz -lbz2
> -lfontconfig -lexpat -lfreetype -lz -lbz2      -lgnutls -lnettle
> -lhogweed -lgmp -lintl -liconv -ltasn1 -lp11-kit -lz   -lpthread
> -lgio-2.0 -lz -lgmodule-2.0 -lgobject-2.0 -lffi -lglib-2.0 -lintl
> -liconv -lpcre
> ../lib/libgnu.a(pipe2.o): In function `rpl_pipe2':
> /c/src/emacs/64testt/lib/pipe2.c:144: undefined reference to `setmode'
>
>
> I think the issue here is that pipe2.c shouldn't be compiled on Cygwin.

Sorry, that was a stupid guess, made without even  looking at pipe2.c. 
But the point is that setmode shouldn't be used on Cygwin (and wasn't 
prior to revision 113315).

The following patch allows the build to complete, but I don't know if 
it's the right fix:

=== modified file 'lib/pipe2.c'
--- lib/pipe2.c 2013-07-07 18:00:14 +0000
+++ lib/pipe2.c 2013-07-09 12:20:47 +0000
@@ -138,7 +138,7 @@
          goto fail;
      }

-# if O_BINARY
+# if O_BINARY && !defined (CYGWIN)
    if (flags & O_BINARY)
      {
        setmode (fd[1], O_BINARY);



Ken

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Ken Brown]

On 7/9/2013 8:33 AM, Ken Brown wrote:
> But the point is that setmode shouldn't be used on Cygwin (and wasn't
> prior to revision 113315).
>
> The following patch allows the build to complete, but I don't know if
> it's the right fix:
>
> === modified file 'lib/pipe2.c'
> --- lib/pipe2.c 2013-07-07 18:00:14 +0000
> +++ lib/pipe2.c 2013-07-09 12:20:47 +0000
> @@ -138,7 +138,7 @@
>           goto fail;
>       }
>
> -# if O_BINARY
> +# if O_BINARY && !defined (CYGWIN)
>     if (flags & O_BINARY)
>       {
>         setmode (fd[1], O_BINARY);

On second thought, instead of the negative '!defined (CYGWIN)', the code 
should probably explicitly list the platforms on which setmode *should* 
be used.  This is how it's done in other uses of setmode.

Ken

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Eli Zaretskii]

> Date: Tue, 09 Jul 2013 08:33:03 -0400
> From: Ken Brown <kbrown@cornell.edu>
> Cc: Paul Eggert <eggert@cs.ucla.edu>
> 
> But the point is that setmode shouldn't be used on Cygwin

Why not?  What if the volume in question was mounted in text mode?
And even if it wasn't, wouldn't setmode be a no-op?

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Ken Brown]

On 7/9/2013 12:17 PM, Eli Zaretskii wrote:
>> Date: Tue, 09 Jul 2013 08:33:03 -0400
>> From: Ken Brown <kbrown@cornell.edu>
>> Cc: Paul Eggert <eggert@cs.ucla.edu>
>>
>> But the point is that setmode shouldn't be used on Cygwin
>
> Why not?  What if the volume in question was mounted in text mode?
> And even if it wasn't, wouldn't setmode be a no-op?

No, the symbol 'setmode' is no longer exported in 64bit Cygwin.  That's 
what caused the build failure.  And the code in pipe2.c that calls 
setmode should never actually get executed on Cygwin anyway because of this:

         int result = pipe2 (fd, flags);
         if (!(result < 0 && errno == ENOSYS))
           {
             have_pipe2_really = 1;
             return result;

Ken

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Eli Zaretskii]

> Date: Tue, 09 Jul 2013 12:46:44 -0400
> From: Ken Brown <kbrown@cornell.edu>
> CC: 14821@debbugs.gnu.org, eggert@cs.ucla.edu
> 
> the symbol 'setmode' is no longer exported in 64bit Cygwin.

Are you saying that there's no way of reading files in text mode while
removing CR characters from CR-LF pairs?

> That's 
> what caused the build failure.  And the code in pipe2.c that calls 
> setmode should never actually get executed on Cygwin anyway because of this:
> 
>          int result = pipe2 (fd, flags);
>          if (!(result < 0 && errno == ENOSYS))
>            {
>              have_pipe2_really = 1;
>              return result;

So define setmode away for Cygwin versions that don't have it.  Not
everyone runs the bleeding edge.

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Ken Brown]

On 7/9/2013 1:07 PM, Eli Zaretskii wrote:
>> Date: Tue, 09 Jul 2013 12:46:44 -0400
>> From: Ken Brown <kbrown@cornell.edu>
>> CC: 14821@debbugs.gnu.org, eggert@cs.ucla.edu
>>
>> the symbol 'setmode' is no longer exported in 64bit Cygwin.
>
> Are you saying that there's no way of reading files in text mode while
> removing CR characters from CR-LF pairs?
>
>> That's
>> what caused the build failure.  And the code in pipe2.c that calls
>> setmode should never actually get executed on Cygwin anyway because of this:
>>
>>           int result = pipe2 (fd, flags);
>>           if (!(result < 0 && errno == ENOSYS))
>>             {
>>               have_pipe2_really = 1;
>>               return result;
>
> So define setmode away for Cygwin versions that don't have it.  Not
> everyone runs the bleeding edge.

The discussion is moot now.  Paul just submitted a patch to Gnulib that 
should fix it in a better way than what I suggested.

Ken

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Paul Eggert]

On 07/09/13 10:13, Ken Brown wrote:
> Paul just submitted a patch to Gnulib that should fix it in a better way than what I suggested.

One problem was that I tried too hard to not use Gnulib
in Emacs, so I mistakenly avoided Gnulib's binary-io module.

There appears to be another issue, though -- a porting bug in Gnulib when
using the accept4 or pipe2 modules on Cygwin.  As Ken notes I
have tried to fix that problem in Gnulib, and propagated the
fix to Emacs while I was fixing the binary-io issue.

Please try trunk bzr 113348 to see whether these two changes
solve the problem.

bug#14821: Build broken on 64-bit Cygwin starting with bzr revision 113315

[Ken Brown]

On 7/9/2013 1:29 PM, Paul Eggert wrote:
> On 07/09/13 10:13, Ken Brown wrote:
>> Paul just submitted a patch to Gnulib that should fix it in a better way than what I suggested.
>
> One problem was that I tried too hard to not use Gnulib
> in Emacs, so I mistakenly avoided Gnulib's binary-io module.
>
> There appears to be another issue, though -- a porting bug in Gnulib when
> using the accept4 or pipe2 modules on Cygwin.  As Ken notes I
> have tried to fix that problem in Gnulib, and propagated the
> fix to Emacs while I was fixing the binary-io issue.
>
> Please try trunk bzr 113348 to see whether these two changes
> solve the problem.

Yes, that fixes it.  Thanks.

For the record, the problem I reported was due to the fact that Cygwin 
defines setmode as a macro that expands to _setmode.  (So I was wrong 
when I said that setmode shouldn't be used on Cygwin.)  But this 
definition is in <io.h>, which wasn't being #included.  The problem 
showed up only in the 64-bit build because underscores are automatically 
prepended in the 32-bit build.

I'm closing the bug.

Ken