From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <49A66CF6.7050405@domain.hid> Date: Thu, 26 Feb 2009 11:20:38 +0100 From: Gilles Chanteperdrix MIME-Version: 1.0 References: <49A4209F.60502@domain.hid> <49A51C3D.1020705@domain.hid> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: [Xenomai-core] [PATCH] blackfin: build xenomai in FLAT format List-Id: "Xenomai life and development \(bug reports, patches, discussions\)" List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: yi li Cc: Xenomai-core@domain.hid yi li wrote: > On Wed, Feb 25, 2009 at 6:23 PM, Gilles Chanteperdrix > wrote: >> yi li wrote: >>> On Wed, Feb 25, 2009 at 12:30 AM, Philippe Gerum wrote: >>>>> 2. libpthread_rt.la should not depend on "lpthread". >>>>> >>>> Nak. In flat mode, turning the link dependencies order upside down will not buy >>>> us anything. Two-phase link is the only way to prevent circular/invalid >>>> dependencies with a static only format when symbol wrapping is involved. >>>> >>> Agree. But "libpthread_rt_la_LDFLAGS" should not include "-lpthread", >>> otherwise libtool will think libpthread_rt.a depends on libpthread. >>> I don't think there is such a dependency, isn't it? >> Look at libpthread_rt code. >> > > Here is the problem: > > /bin/bash ../../../libtool --tag=CC --mode=link bfin-uclinux-gcc > -pipe -Wall -g -O2 -mcpu=bf527-0.1 > -Wl,@/home/adam/new_workspace/local_svn/kernel/adeos/uclinux-dist/user/xenomai/xenomai-2.4.x/src/skins/posix/posix.wrappers > -D_GNU_SOURCE -D_REENTRANT -Wall -pipe -Wl,-elf2flt -mcpu=bf527-0.1 > -o cyclictest cyclictest-cyclictest.o -lpthread -lrt > ../../skins/posix/libpthread_rt.la > mkdir .libs > bfin-uclinux-gcc -pipe -Wall -g -O2 -mcpu=bf527-0.1 > -Wl,@/home/adam/new_workspace/local_svn/kernel/adeos/uclinux-dist/user/xenomai/xenomai-2.4.x/src/skins/posix/posix.wrappers > -D_GNU_SOURCE -D_REENTRANT -Wall -pipe -Wl,-elf2flt -mcpu=bf527-0.1 -o > cyclictest cyclictest-cyclictest.o -lrt > ../../skins/posix/.libs/libpthread_rt.a -lpthread > /home/adam/workspace/toolchain/4.1-1220/bfin-uclinux/bin/../bfin-uclinux/runtime/usr/lib/libpthread.a(manager.o): > In function `__pthread_manager': > libpthread/linuxthreads.old/manager.c:173: undefined reference to > `___wrap_select' > /home/adam/workspace/toolchain/4.1-1220/bfin-uclinux/bin/../bfin-uclinux/runtime/usr/lib/libpthread.a(mutex.o): > In function `__pthread_once_fork_child': > libpthread/linuxthreads.old/mutex.c:357: undefined reference to > `___wrap_pthread_cond_init' > /home/adam/workspace/toolchain/4.1-1220/bfin-uclinux/bin/../bfin-uclinux/runtime/usr/lib/libpthread.a(mutex.o): > In function `__pthread_once': > libpthread/linuxthreads.old/mutex.c:308: undefined reference to > `___wrap_pthread_cond_wait' > libpthread/linuxthreads.old/mutex.c:325: undefined reference to > `___wrap_pthread_cond_broadcast' > /home/adam/workspace/toolchain/4.1-1220/bfin-uclinux/bin/../bfin-uclinux/runtime/usr/lib/libpthread.a(mutex.o): > In function `pthread_once_cancelhandler': > libpthread/linuxthreads.old/mutex.c:281: undefined reference to > `___wrap_pthread_cond_broadcast' > collect2: ld returned 1 exit status > > In my opinion, "-lpthread" does not need to be set in > "libpthread_rt_la_LDFLAGS". The application (e.g, cyclictest) decides > which libraries to link with. -lpthread does need to be set in libpthread_rt_la_LDFLAGS, if you look at libpthread_rt code (which you apparently do not want to dot), you will see that it uses libpthread functions. The issue you have has already been discussed on this list, and we already know the correct workaround. The correct workaround is not to link with -lpthread first. Linking with -lpthread first leads to broken binaries, whose libpthread functions call libpthread_rt functions. The correct workaround is to link your binary in two stages, or to forget about FLAT and use FDPIC. -- Gilles.