RE: PATCH: Fix bfd.c (Re: Tracking down a build breakage)

RE: PATCH: Fix bfd.c (Re: Tracking down a build breakage)

[Kazu Hirata]

Hi,

> But my bfd_sprintf_vma/bfd_fprintf_vma change does break any BFD
> which doesn't include elf.lo. It seems to me that we should only
> check bfd_target_elf_flavour when elf is configured in BFD. Here
> is a patch. Any comments?

This seemd to be a problem when I was trying to build gdb.

> 2001-08-11  H.J. Lu  <hjl@gnu.org>
> 
> 	* Makefile.am (BFD_DEFINES): New. Substituted by @bfd_defines@.
> 	(INCLUDES): Add $(BFD_DEFINES).
> 	* Makefile.in: Regenerated.
> 
> 	* configure.in (bfd_defines): Add -DHAVE_BFD_ELF if elf.lo
> 	is in $bfd_backend.
> 	(AC_SUBST(bfd_defines)): Added.
> 	* configure: Regenerated.
> 
> 	* bfd.c (bfd_sprintf_vma): Check ELF only if HAVE_BFD_ELF is
> 	defined.
> 	(bfd_fprintf_vma): Likewise.

Amazingly, after doing "cvs co binutils", I did not have to apply your
patch to successfully build h8300 target.  With your patch, the build
process got stuck as follows.  The command line looks just fine.  I
don't know why gcc makes such a complaint...

/bin/sh ./libtool --mode=compile gcc -DHAVE_CONFIG_H -I. -I/home/kazu/gnu/binutils/binutils-head/src/bfd -I. -D_GNU_SOURCE @bfd_defines@ -I. -I/home/kazu/gnu/binutils/binutils-head/src/bfd -I/home/kazu/gnu/binutils/binutils-head/src/bfd/../include  -I/home/kazu/gnu/binutils/binutils-head/src/bfd/../intl -I../intl   -W -Wall -Wstrict-prototypes -Wmissing-prototypes -g -O2 -c /home/kazu/gnu/binutils/binutils-head/src/bfd/archive.c
gcc -DHAVE_CONFIG_H -I. -I/home/kazu/gnu/binutils/binutils-head/src/bfd -I. -D_GNU_SOURCE @bfd_defines@ -I. -I/home/kazu/gnu/binutils/binutils-head/src/bfd -I/home/kazu/gnu/binutils/binutils-head/src/bfd/../include -I/home/kazu/gnu/binutils/binutils-head/src/bfd/../intl -I../intl -W -Wall -Wstrict-prototypes -Wmissing-prototypes -g -O2 -c /home/kazu/gnu/binutils/binutils-head/src/bfd/archive.c -o archive.o
gcc: cannot specify -o with -c or -S and multiple compilations
make[3]: *** [archive.lo] Error 1
make[3]: Leaving directory `/var/tmp/h8300-hms-binutils-head-0811-2/bfd'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/var/tmp/h8300-hms-binutils-head-0811-2/bfd'
make[1]: *** [all-recursive-am] Error 2
make[1]: Leaving directory `/var/tmp/h8300-hms-binutils-head-0811-2/bfd'
make: *** [all-bfd] Error 2
Command exited with non-zero status 2

Kazu Hirata

Re: PATCH: Fix bfd.c (Re: Tracking down a build breakage)

[Hans-Peter Nilsson]

On Sun, 12 Aug 2001, Hans-Peter Nilsson wrote:
> I don't think this is fully fixed.  Please consider building
> h8300-hms and i960-coff. They don't built at the moment

Whoops.  Forget about that.  Pilot error.  Sorry for wasting
your time.

brgds, H-P

Re: PATCH: Fix bfd.c (Re: Tracking down a build breakage)

[Hans-Peter Nilsson]

On 11 Aug 2001, Nick Clifton wrote:
> > But my bfd_sprintf_vma/bfd_fprintf_vma change does break any BFD
> > which doesn't include elf.lo. It seems to me that we should only
> > check bfd_target_elf_flavour when elf is configured in BFD. Here
> > is a patch. Any comments?
>
> Ooops - I discovered this and checked in a different patch to fix
> this, without actually checking the mailing lists first.  Sorry!

I don't think this is fully fixed.  Please consider building
h8300-hms and i960-coff. They don't built at the moment
complaining that (mind the wrapped lines):

gcc -W -Wall -Wstrict-prototypes -Wmissing-prototypes -g -O2 -o
size size.o bucomm.o version.o filemode.o  ../bfd/.libs/libbfd.a
\
../libiberty/libiberty.a
../bfd/.libs/libbfd.a(bfd.o): In function `bfd_sprintf_vma':
/home/hp/cvs_areas/combined/cvs_write/bfd/bfd.c:1273: undefined
reference to `bfd_elf_sprintf_vma'
../bfd/.libs/libbfd.a(bfd.o): In function `bfd_fprintf_vma':
/home/hp/cvs_areas/combined/cvs_write/bfd/bfd.c:1285: undefined
reference to `bfd_elf_fprintf_vma'
collect2: ld returned 1 exit status
make[3]: *** [size] Error 1

I haven't looked into it further than this.

brgds, H-P

Re: PATCH: Fix bfd.c (Re: Tracking down a build breakage)

[H . J . Lu]

On Sat, Aug 11, 2001 at 11:01:14PM +0100, Nick Clifton wrote:
> Hi H.J.
> 
> > But my bfd_sprintf_vma/bfd_fprintf_vma change does break any BFD
> > which doesn't include elf.lo. It seems to me that we should only
> > check bfd_target_elf_flavour when elf is configured in BFD. Here
> > is a patch. Any comments?
> 
> Ooops - I discovered this and checked in a different patch to fix
> this, without actually checking the mailing lists first.  Sorry!
> 
> Do you have any objections to the solution I used (adding new fields
> to the elf backend data structure) ?
> 

No objections here.


H.J.

Re: PATCH: Fix bfd.c (Re: Tracking down a build breakage)

[Nick Clifton]

Hi H.J.

> But my bfd_sprintf_vma/bfd_fprintf_vma change does break any BFD
> which doesn't include elf.lo. It seems to me that we should only
> check bfd_target_elf_flavour when elf is configured in BFD. Here
> is a patch. Any comments?

Ooops - I discovered this and checked in a different patch to fix
this, without actually checking the mailing lists first.  Sorry!

Do you have any objections to the solution I used (adding new fields
to the elf backend data structure) ?

Cheers
        Nick

Re: PATCH: Fix bfd.c (Re: Tracking down a build breakage)

[H . J . Lu]

On Sat, Aug 11, 2001 at 02:13:28PM -0400, Kazu Hirata wrote:
> Hi,
> 
> > But my bfd_sprintf_vma/bfd_fprintf_vma change does break any BFD
> > which doesn't include elf.lo. It seems to me that we should only
> > check bfd_target_elf_flavour when elf is configured in BFD. Here
> > is a patch. Any comments?
> 
> This seemd to be a problem when I was trying to build gdb.
> 
> > 2001-08-11  H.J. Lu  <hjl@gnu.org>
> > 
> > 	* Makefile.am (BFD_DEFINES): New. Substituted by @bfd_defines@.
> > 	(INCLUDES): Add $(BFD_DEFINES).
> > 	* Makefile.in: Regenerated.
> > 
> > 	* configure.in (bfd_defines): Add -DHAVE_BFD_ELF if elf.lo
> > 	is in $bfd_backend.
> > 	(AC_SUBST(bfd_defines)): Added.
> > 	* configure: Regenerated.
> > 
> > 	* bfd.c (bfd_sprintf_vma): Check ELF only if HAVE_BFD_ELF is
> > 	defined.
> > 	(bfd_fprintf_vma): Likewise.
> 
> Amazingly, after doing "cvs co binutils", I did not have to apply your
> patch to successfully build h8300 target.  With your patch, the build
> process got stuck as follows.  The command line looks just fine.  I
> don't know why gcc makes such a complaint...

I didn't include the patch for bfd/configure. Do

# cd ..../bfd
# autoconf


H.J.

PATCH: Fix bfd.c (Re: Tracking down a build breakage)

[H . J . Lu]

On Sat, Aug 11, 2001 at 12:44:11PM -0400, Kazu Hirata wrote:
> Hi,
> 
> I recently reported to bug-binutils that the h8300 port had a build
> breakage.
> 
> http://sources.redhat.com/ml/bug-binutils/2001-q3/msg00057.html
> 

You have an incomplete source. I use

# cvs co binutils

to update my CVS tree.

But my bfd_sprintf_vma/bfd_fprintf_vma change does break any BFD
which doesn't include elf.lo. It seems to me that we should only
check bfd_target_elf_flavour when elf is configured in BFD. Here
is a patch. Any comments?


H.J.
-----
2001-08-11  H.J. Lu  <hjl@gnu.org>

	* Makefile.am (BFD_DEFINES): New. Substituted by @bfd_defines@.
	(INCLUDES): Add $(BFD_DEFINES).
	* Makefile.in: Regenerated.

	* configure.in (bfd_defines): Add -DHAVE_BFD_ELF if elf.lo
	is in $bfd_backend.
	(AC_SUBST(bfd_defines)): Added.
	* configure: Regenerated.

	* bfd.c (bfd_sprintf_vma): Check ELF only if HAVE_BFD_ELF is
	defined.
	(bfd_fprintf_vma): Likewise.

Index: bfd.c
===================================================================
RCS file: /work/cvs/gnu/binutils/bfd/bfd.c,v
retrieving revision 1.1.1.13
diff -u -p -r1.1.1.13 bfd.c
--- bfd.c	2001/08/10 18:46:20	1.1.1.13
+++ bfd.c	2001/08/11 17:26:22
@@ -1265,24 +1265,28 @@ bfd_record_phdr (abfd, type, flags_valid
 
 void
 bfd_sprintf_vma (abfd, buf, value)
-     bfd *abfd;
+     bfd *abfd ATTRIBUTE_UNUSED;
      char *buf;
      bfd_vma value;
 {
+#ifdef HAVE_BFD_ELF
   if (bfd_get_flavour (abfd) == bfd_target_elf_flavour)
     bfd_elf_sprintf_vma (abfd, buf, value);
   else
+#endif
     sprintf_vma (buf, value);
 }
 
 void
 bfd_fprintf_vma (abfd, stream, value)
-     bfd *abfd;
+     bfd *abfd ATTRIBUTE_UNUSED;
      PTR stream;
      bfd_vma value;
 {
+#ifdef HAVE_BFD_ELF
   if (bfd_get_flavour (abfd) == bfd_target_elf_flavour)
     bfd_elf_fprintf_vma (abfd, stream, value);
   else
+#endif
     fprintf_vma ((FILE *) stream, value);
 }
Index: Makefile.am
===================================================================
RCS file: /work/cvs/gnu/binutils/bfd/Makefile.am,v
retrieving revision 1.47
diff -u -p -r1.47 Makefile.am
--- Makefile.am	2001/07/24 16:33:47	1.47
+++ Makefile.am	2001/08/11 17:26:23
@@ -475,9 +475,12 @@ WORDSIZE = @wordsize@
 ALL_BACKENDS = @all_backends@
 BFD_BACKENDS = @bfd_backends@
 BFD_MACHINES = @bfd_machines@
+BFD_DEFINES = @bfd_defines@
 TDEFAULTS = @tdefaults@
 
-INCLUDES = -D_GNU_SOURCE @HDEFINES@ @COREFLAG@ @TDEFINES@ $(CSEARCH) $(CSWITCHES) -I$(srcdir)/../intl -I../intl
+INCLUDES = -D_GNU_SOURCE @HDEFINES@ @COREFLAG@ @TDEFINES@ \
+	   $(BFD_DEFINES) $(CSEARCH) $(CSWITCHES) \
+	   -I$(srcdir)/../intl -I../intl
 
 # C source files that correspond to .o's.
 SOURCE_CFILES = \
Index: Makefile.in
===================================================================
RCS file: /work/cvs/gnu/binutils/bfd/Makefile.in,v
retrieving revision 1.54
diff -u -p -r1.54 Makefile.in
--- Makefile.in	2001/07/24 16:33:48	1.54
+++ Makefile.in	2001/08/11 17:26:23
@@ -114,6 +114,7 @@ WIN32LDFLAGS = @WIN32LDFLAGS@
 WIN32LIBADD = @WIN32LIBADD@
 all_backends = @all_backends@
 bfd_backends = @bfd_backends@
+bfd_defines = @bfd_defines@
 bfd_machines = @bfd_machines@
 l = @l@
 tdefaults = @tdefaults@
@@ -604,10 +605,14 @@ WORDSIZE = @wordsize@
 ALL_BACKENDS = @all_backends@
 BFD_BACKENDS = @bfd_backends@
 BFD_MACHINES = @bfd_machines@
+BFD_DEFINES = @bfd_defines@
 TDEFAULTS = @tdefaults@
 
-INCLUDES = -D_GNU_SOURCE @HDEFINES@ @COREFLAG@ @TDEFINES@ $(CSEARCH) $(CSWITCHES) -I$(srcdir)/../intl -I../intl
+INCLUDES = -D_GNU_SOURCE @HDEFINES@ @COREFLAG@ @TDEFINES@ \
+	   $(BFD_DEFINES) $(CSEARCH) $(CSWITCHES) \
+	   -I$(srcdir)/../intl -I../intl
 
+
 # C source files that correspond to .o's.
 SOURCE_CFILES = \
 	$(BFD_LIBS_CFILES) \
Index: configure.in
===================================================================
RCS file: /work/cvs/gnu/binutils/bfd/configure.in,v
retrieving revision 1.165
diff -u -p -r1.165 configure.in
--- configure.in	2001/08/10 17:03:21	1.165
+++ configure.in	2001/08/11 17:26:23
@@ -128,6 +128,8 @@ BFD_NEED_DECLARATION(realloc)
 BFD_NEED_DECLARATION(free)
 BFD_NEED_DECLARATION(getenv)
 
+bfd_defines=
+
 # If we are configured native, pick a core file support file.
 COREFILE=
 COREFLAG=
@@ -702,6 +704,12 @@ ta="$f"
 bfd_backends="$tb"
 bfd_machines="$ta"
 
+case "$bfd_backends" in
+*elf.lo*)
+  bfd_defines="$bfd_defines -DHAVE_BFD_ELF"
+  ;;
+esac
+
 if test x${all_targets} = xtrue ; then
   bfd_backends="${bfd_backends}"' $(ALL_BACKENDS)'
   bfd_machines="${bfd_machines}"' $(ALL_MACHINES)'
@@ -738,6 +746,7 @@ AC_SUBST(wordsize)
 AC_SUBST(all_backends)
 AC_SUBST(bfd_backends)
 AC_SUBST(bfd_machines)
+AC_SUBST(bfd_defines)
 
 tdefaults=""
 test -n "${defvec}" && tdefaults="${tdefaults} -DDEFAULT_VECTOR=${defvec}"