[RFA] Allow binutils compilation for mingw64 with --enable-targets=all

[RFA] Allow binutils compilation for mingw64 with --enable-targets=all

[Pierre Muller]

Trying to compile Binutils with --enable-targets=all
for mingw64 leads to this error:
x86_64-w64-mingw32-gcc -DHAVE_CONFIG_H -I. -I../../src/binutils  -I.
-I../../src
/binutils -I../bfd -I../../src/binutils/../bfd
-I../../src/binutils/../include -
I./../intl -DLOCALEDIR="\"/usr/local/share/locale\""
-Dbin_dummy_emulation=bin_v
anilla_emulation  -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wshadow
-Wn
o-format -Werror -gstabs+ -O0 -D__USE_MINGW_ACCESS -MT coffdump.o -MD -MP
-MF .d
eps/coffdump.Tpo -c -o coffdump.o ../../src/binutils/coffdump.c
cc1: warnings being treated as errors
../../src/binutils/coffdump.c: In function 'dump_coff_scope':
../../src/binutils/coffdump.c:368:37: error: cast from pointer to integer of
dif
ferent size
make[2]: *** [coffdump.o] Error 1


The patch below fixes that compilation error.



Pierre Muller
GDB pascal language maintainer

2011-03-25  Pierre Muller  <muller@ics.u-strasbg.fr>

	* coffdump.c (dump_coff_scope): Use double typecast for pointer P
	to allow compilation for all targets.

Index: coffdump.c
===================================================================
RCS file: /cvs/src/src/binutils/coffdump.c,v
retrieving revision 1.18
diff -u -p -r1.18 coffdump.c
--- coffdump.c	5 Jul 2007 16:54:45 -0000	1.18
+++ coffdump.c	25 Mar 2011 21:34:07 -0000
@@ -365,7 +365,7 @@ dump_coff_scope (struct coff_scope *p)
   if (p)
     {
       tab (1);
-      printf ("List of blocks %lx ",(unsigned long) p);
+      printf ("List of blocks %" BFD_VMA_FMT "x ",(bfd_vma) (uintptr_t) p);
 
       if (p->sec)
 	printf( "  %s %x..%x",  p->sec->name,p->offset, p->offset + p->size
-1);

Re: [RFA] Allow binutils compilation for mingw64 with --enable-targets=all

[Alan Modra]

On Fri, Mar 25, 2011 at 10:51:35PM +0100, Pierre Muller wrote:
> 	* coffdump.c (dump_coff_scope): Use double typecast for pointer P
> 	to allow compilation for all targets.

OK, thanks!

-- 
Alan Modra
Australia Development Lab, IBM

[RFA] Minor fixes to coffdump.c source

[Pierre Muller]

> -----Message d'origine-----
> De : binutils-owner@sourceware.org [mailto:binutils-owner@sourceware.org]
De
> la part de Alan Modra
> Envoyé : lundi 28 mars 2011 12:27
> À : Pierre Muller
> Cc : 'Binutils'
> Objet : Re: [RFA] Allow binutils compilation for mingw64 with --enable-
> targets=all
> 
> On Fri, Mar 25, 2011 at 10:51:35PM +0100, Pierre Muller wrote:
> > 	* coffdump.c (dump_coff_scope): Use double typecast for pointer P
> > 	to allow compilation for all targets.
> 
> OK, thanks!

  Thanks for the approval,
patch committed.

  While doing this, I noticed two mistakes:

May I commit those also?

Pierre

2011-03-28  Pierre Muller  <muller@ics.u-strasbg.fr>

	* coffdump.c (coff_dump): Correct spelling error.
	(show_usage): Replace SYSROFF by COFF.

Index: coffdump.c
===================================================================
RCS file: /cvs/src/src/binutils/coffdump.c,v
retrieving revision 1.19
diff -u -p -r1.19 coffdump.c
--- coffdump.c	28 Mar 2011 11:58:18 -0000	1.19
+++ coffdump.c	28 Mar 2011 12:06:19 -0000
@@ -444,7 +444,7 @@ coff_dump (struct coff_ofile *ptr)
 
   printf ("Coff dump");
   nl ();
-  printf ("#souces %d", ptr->nsources);
+  printf ("#sources %d", ptr->nsources);
   nl ();
   dump_coff_sfile (ptr->source_head);
 
@@ -458,7 +458,7 @@ static void
 show_usage (FILE *file, int status)
 {
   fprintf (file, _("Usage: %s [option(s)] in-file\n"), program_name);
-  fprintf (file, _(" Print a human readable interpretation of a SYSROFF
object file\n"));
+  fprintf (file, _(" Print a human readable interpretation of a COFF object
file\n"));
   fprintf (file, _(" The options are:\n\
   @<file>                Read options from <file>\n\
   -h --help              Display this information\n\

Re: [RFA] Minor fixes to coffdump.c source

[Alan Modra]

On Mon, Mar 28, 2011 at 02:10:07PM +0200, Pierre Muller wrote:
> 	* coffdump.c (coff_dump): Correct spelling error.
> 	(show_usage): Replace SYSROFF by COFF.

Please commit.

-- 
Alan Modra
Australia Development Lab, IBM

RE: [RFA] Minor fixes to coffdump.c source

[Pierre Muller]

> -----Message d'origine-----
> De : Alan Modra [mailto:amodra@gmail.com]
> Envoyé : lundi 28 mars 2011 17:06
> À : Pierre Muller
> Cc : 'Binutils'
> Objet : Re: [RFA] Minor fixes to coffdump.c source
> 
> On Mon, Mar 28, 2011 at 02:10:07PM +0200, Pierre Muller wrote:
> > 	* coffdump.c (coff_dump): Correct spelling error.
> > 	(show_usage): Replace SYSROFF by COFF.
> 
> Please commit.

  Thank you for the approval.
Patch cheked in.

Pierre Muller

Re: [RFA] Allow binutils compilation for mingw64 with --enable-targets=all

[Alan Modra]

On Fri, Mar 25, 2011 at 10:51:35PM +0100, Pierre Muller wrote:
> 	* coffdump.c (dump_coff_scope): Use double typecast for pointer P
> 	to allow compilation for all targets.

I missed this when reviewing.  For systems that don't have uintptr_t.

	* coffdump.c: Include bfd_stdint.h

Index: binutils/coffdump.c
===================================================================
RCS file: /cvs/src/src/binutils/coffdump.c,v
retrieving revision 1.20
diff -u -p -r1.20 coffdump.c
--- binutils/coffdump.c	28 Mar 2011 15:24:02 -0000	1.20
+++ binutils/coffdump.c	29 Mar 2011 02:05:02 -0000
@@ -1,6 +1,6 @@
 /* Coff file dumper.
-   Copyright 1994, 1995, 1998, 1999, 2000, 2001, 2002, 2003, 2005, 2007
-   Free Software Foundation, Inc.
+   Copyright 1994, 1995, 1998, 1999, 2000, 2001, 2002, 2003, 2005, 2007,
+   2011 Free Software Foundation, Inc.
 
    This file is part of GNU Binutils.
 
@@ -27,6 +27,7 @@
 
 #include "sysdep.h"
 #include "bfd.h"
+#include "bfd_stdint.h"
 #include "libiberty.h"
 #include "bucomm.h"
 
-- 
Alan Modra
Australia Development Lab, IBM

RE: [RFA] Allow binutils compilation for mingw64 with --enable-targets=all

[Pierre Muller]

> > -      printf ("List of blocks %lx ",(unsigned long) p);
> > +      printf ("List of blocks %" BFD_VMA_FMT "x ",(bfd_vma) (uintptr_t)
> p);
> I can't approve this. Yes, thanks for catching this.  But is it really
> necessary to do the uintptr_t cast before the final bfd_vma cast?
> bfd_vma should be always >= sizeof (void *), isn't it?

  I think that you get a warning as soon as the
size is different, and typically on a 32-bit system
with 64-bit bfd enabled, bfd_vma will be 
a 64-bit integer and p a 32-bit pointer, which will generate
a warning that is suppressed by this double typecast.

Pierre Muller

Re: [RFA] Allow binutils compilation for mingw64 with --enable-targets=all

[Kai Tietz]

2011/3/25 Pierre Muller <pierre.muller@ics-cnrs.unistra.fr>:
> Trying to compile Binutils with --enable-targets=all
> for mingw64 leads to this error:
> x86_64-w64-mingw32-gcc -DHAVE_CONFIG_H -I. -I../../src/binutils  -I.
> -I../../src
> /binutils -I../bfd -I../../src/binutils/../bfd
> -I../../src/binutils/../include -
> I./../intl -DLOCALEDIR="\"/usr/local/share/locale\""
> -Dbin_dummy_emulation=bin_v
> anilla_emulation  -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wshadow
> -Wn
> o-format -Werror -gstabs+ -O0 -D__USE_MINGW_ACCESS -MT coffdump.o -MD -MP
> -MF .d
> eps/coffdump.Tpo -c -o coffdump.o ../../src/binutils/coffdump.c
> cc1: warnings being treated as errors
> ../../src/binutils/coffdump.c: In function 'dump_coff_scope':
> ../../src/binutils/coffdump.c:368:37: error: cast from pointer to integer of
> dif
> ferent size
> make[2]: *** [coffdump.o] Error 1
>
>
> The patch below fixes that compilation error.
>
>
>
> Pierre Muller
> GDB pascal language maintainer
>
> 2011-03-25  Pierre Muller  <muller@ics.u-strasbg.fr>
>
>        * coffdump.c (dump_coff_scope): Use double typecast for pointer P
>        to allow compilation for all targets.
>
> Index: coffdump.c
> ===================================================================
> RCS file: /cvs/src/src/binutils/coffdump.c,v
> retrieving revision 1.18
> diff -u -p -r1.18 coffdump.c
> --- coffdump.c  5 Jul 2007 16:54:45 -0000       1.18
> +++ coffdump.c  25 Mar 2011 21:34:07 -0000
> @@ -365,7 +365,7 @@ dump_coff_scope (struct coff_scope *p)
>   if (p)
>     {
>       tab (1);
> -      printf ("List of blocks %lx ",(unsigned long) p);
> +      printf ("List of blocks %" BFD_VMA_FMT "x ",(bfd_vma) (uintptr_t) p);
>
>       if (p->sec)
>        printf( "  %s %x..%x",  p->sec->name,p->offset, p->offset + p->size
> -1);
>
>

I can't approve this. Yes, thanks for catching this.  But is it really
necessary to do the uintptr_t cast before the final bfd_vma cast?
bfd_vma should be always >= sizeof (void *), isn't it?

Kai