[Bug boehm-gc/46311] New: boehm-gc build problem with uclibc

[Bug boehm-gc/46311] New: boehm-gc build problem with uclibc

[dougmencken at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46311

           Summary: boehm-gc build problem with uclibc
           Product: gcc
           Version: 4.5.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: boehm-gc
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: dougmencken@gmail.com


I'm trying to get GCC with GCJ (and related libs). The ./configure options I
use are the following:

--------------------------------------------------------------------------------
../"${SOURCE_DIR}"/configure --prefix=/usr --sysconfdir=/etc
--mandir=/usr/share/man \
--enable-languages=c,c++,fortran,java --with-cpu=default32 \
--with-system-zlib --enable-threads=posix \
--disable-nls --disable-werror --disable-multilib --disable-libssp \
--disable-__cxa_atexit --disable-checking --enable-shared \
--enable-libgcj --enable-libgcj-multifile --enable-interpreter
--disable-libjava-multilib
--------------------------------------------------------------------------------

At stage 3, make fails with the following error:

--------------------------------------------------------------------------------
/bin/bash ./libtool   --mode=compile /root/build-farm/_gcc_bootstrap/./gcc/xgcc
-B/root/build-farm/_gcc_bootstrap/./gcc/ -B/usr/powerpc-gnu-linux-uclibc/bin/
-B/usr/powerpc-gnu-linux-uclibc/lib/ -isystem
/usr/powerpc-gnu-linux-uclibc/include -isystem
/usr/powerpc-gnu-linux-uclibc/sys-include    -DHAVE_CONFIG_H
-I/root/build-farm/gcc-v4.5.1.sourcedir/boehm-gc/include  -fexceptions
-Iinclude -I././targ-include -I.//libc/include -g -O2 -fexceptions -Iinclude
-I././targ-include -I.//libc/include  -c -o os_dep.lo
../../../gcc-v4.5.1.sourcedir/boehm-gc/os_dep.c
libtool: compile:  /root/build-farm/_gcc_bootstrap/./gcc/xgcc
-B/root/build-farm/_gcc_bootstrap/./gcc/ -B/usr/powerpc-gnu-linux-uclibc/bin/
-B/usr/powerpc-gnu-linux-uclibc/lib/ -isystem
/usr/powerpc-gnu-linux-uclibc/include -isystem
/usr/powerpc-gnu-linux-uclibc/sys-include -DHAVE_CONFIG_H
-I/root/build-farm/gcc-v4.5.1.sourcedir/boehm-gc/include -fexceptions -Iinclude
-I././targ-include -I.//libc/include -g -O2 -fexceptions -Iinclude
-I././targ-include -I.//libc/include -c
../../../gcc-v4.5.1.sourcedir/boehm-gc/os_dep.c  -fPIC -DPIC -o .libs/os_dep.o
../../../gcc-v4.5.1.sourcedir/boehm-gc/os_dep.c:4025:23: fatal error:
execinfo.h: No such file or directory
compilation terminated.
make[3]: *** [os_dep.lo] Error 1
make[3]: Leaving directory
`/root/build-farm/_gcc_bootstrap/powerpc-gnu-linux-uclibc/boehm-gc'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory
`/root/build-farm/_gcc_bootstrap/powerpc-gnu-linux-uclibc/boehm-gc'
make[1]: *** [all-target-boehm-gc] Error 2
make[1]: Leaving directory `/root/build-farm/_gcc_bootstrap'
make: *** [all] Error 2
--------------------------------------------------------------------------------

Diving into that problem, I figured out that boehm-gc sources (actually,
./boehm-gc/include/gc.h) checks if __GLIBC__ is defined, and then trying to
include execinfo.h.

Back in 2006, Solar from Gentoo team already wrote a patch to fix that issue:
http://bugs.gentoo.org/121463 (the patch itself:
http://bugs.gentoo.org/attachment.cgi?id=80702 ). With this patch, boehm-gc
builds without errors.

[Bug boehm-gc/46311] boehm-gc build problem with uclibc

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46311

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |Hans.Boehm at hp dot com

--- Comment #1 from David Edelsohn <dje at gcc dot gnu.org> 2010-11-07 19:57:00 UTC ---
Boehm-GC bugs best are handled upstream by the Boehm-GC community.

[Bug boehm-gc/46311] boehm-gc build problem with uclibc

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46311

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |MOVED

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-02-01 02:44:23 UTC ---
__GLIBC__ should not be defined for uclibc so this is/was an uclibc issue.

[Bug boehm-gc/46311] boehm-gc build problem with uclibc

[dougmencken at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46311

Douglas Mencken <dougmencken at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|MOVED                       |

--- Comment #3 from Douglas Mencken <dougmencken at gmail dot com> 2012-02-01 19:04:19 UTC ---
Re-opening this. uClibc can't stop defining __GLIBC__ due to compatibility with
a lot of stuff, especially ones which use autotools. 

I still must apply this patch:
ftp://ftp.osuosl.org/pub/manulix/scripts/build-scripts/PPFILES/ppfiles-gcc/boehm-gc_uclibc_gc.h.patch
to get the current GCC to be built with GCJ.

Also, by the way, in bdw-gc, starting from the
http://www.hpl.hp.com/personal/Hans_Boehm/gc/gc_source/bdwgc-7_2alpha5-20110107.tar.bz2
release, this issue has been already fixed.

[Bug boehm-gc/46311] boehm-gc build problem with uclibc

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46311

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-02-01 19:51:58 UTC ---
This is a bug in uclibc defining __GLIBC__ and yes they partly get away with it
but that is still the bug.  Maybe if they define __GLIBC__ they should
implement all the features of glibc including execinfo.h .  This is why this is
not a beohm-gc issue but rather an uclibc issue.