Am Dienstag, 29. November 2005 21.32 schrieb Philippe Gerum: > Niklaus Giger wrote: > > Hi > > > > I update to the latest xenomai/trunk (revision 207) and compiled and > > installed without problems xenomai for my PPC405 board (using busybox). > > > > The kernel starts (using a NFS root) and I can login without major > > problems. Tools like xeno-load and xeno-test do not work as there is no > > /bin/bash around. I usually work around this problem by calling insmod > > and alike manually. > > > > In xeno-load I also remarked that > > prefix="/mnt/data.ng/hcu/rootfs/usr/xenomai" > > points to the mount point while compiling/installing xenomai, but not to > > the mount point after starting the target (where / > > equals /mnt/data.ng/hcu/rootfs). > > The proper way to use a staging directory is: > > blah/configure --prefix=/your/target/root > make DESTDIR=/mnt/host/dir > > So that files are installed to /mnt/host/dir/your/target/root with the > target prefix info still being /your/target/root. > > Is this the way you compile/install Xeno? No, I tried --prefix=/your/target/root. But I will fix this soon > > > My real problem is that I cannot link my custom UVM application, as I get > > the following error. Here are as the complete calls to compile (only the > > last file) and link the application: > > > > ccache > > /home/hcu/tools/powerpc-405-linux-gnu/gcc-3.4.1-glibc-2.3.3/bin/powerpc-4 > >05-linux-gnu-g++ -c -Wno-deprecated -g -DAB_DSP_5 -DVXWORKS -DBB_VERS > > -DIMC_VERS -DSYM_VERS -DCEL_VERS -DRTAI -I. > > -I/mnt/data.ng/hcu/rootfs/usr/xenomai/include -O2 -D_GNU_SOURCE > > -D_REENTRANT -D__XENO__ -Wall -pipe -D__XENO_UVM__ -O > > -I/home/hcu/project/metaframe/1_3_5_1/BSys/mak/linux > > -I/home/hcu/project/metaframe/1_3_5_1/BSys/src > > -I/home/hcu/project/metaframe/1_3_5_1 -I/home/hcu/project/vx_skin > > -I/mnt/data.ng/hcu/rootfs/usr/xenomai/include/vxworks > > -I/mnt/data.ng/hcu/rootfs/usr/xenomai/include > > -I/mnt/data.ng/hcu/kernel/build-xeno-hcu3/src/include > > -I/mnt/data.ng/hcu/rootfs/usr/xenomai/include/vxworks > > -O > > /home/hcu/project/metaframe/1_3_5_1/BSys/src/BSysTypeInfoVXWorksParser.cp > >p -o ../../obj/xeno-ppc405/BSysTypeInfoVXWorksParser.o > > ar ru ../../obj/xeno-ppc405/BSys_lib.a \ > > ../../obj/xeno-ppc405/BSysTypeInfoVXWorksParser.o > > ranlib ../../obj/xeno-ppc405/BSys_lib.a > > /home/hcu/tools/powerpc-405-linux-gnu/gcc-3.4.1-glibc-2.3.3/bin/powerpc-4 > >05-linux-gnu-g++ -g -o > > TSysMain ../../obj/xeno-ppc405/TSysMain.o > > ../../obj/xeno-ppc405/BSys_test.a ../../obj/xeno-ppc405/BSys_lib.a > > -u__xeno_skin_init -L/mnt/data.ng/hcu/rootfs/usr/xenomai/lib -luvm > > -lnucleus -lpthread -lvxworks -lvxworks > > > > /mnt/data.ng/hcu/rootfs/usr/xenomai/lib/libnucleus.a(pod.o)(.text+0x674): > > In > > > > function `xnpod_init': > > : undefined reference to `xnarch_atomic_set' > > > > /mnt/data.ng/hcu/rootfs/usr/xenomai/lib/libnucleus.a(pod.o)(.text+0x6b8): > > In > > > > function `xnpod_init': > > : undefined reference to `xnarch_atomic_set' > > > > /mnt/data.ng/hcu/rootfs/usr/xenomai/lib/libnucleus.a(pod.o)(.text+0xfe8): > > In > > > > function `xnpod_set_thread_mode': > > : undefined reference to `xnarch_atomic_set' > > > > /mnt/data.ng/hcu/rootfs/usr/xenomai/lib/libnucleus.a(pod.o)(.text+0x26a8) > >: In > > > > function `xnpod_delete_thread': > > : undefined reference to `xnarch_atomic_set' > > > > /mnt/data.ng/hcu/rootfs/usr/xenomai/lib/libnucleus.a(pod.o)(.text+0x29a8) > >: In > > > > function `xnpod_restart_thread': > > : undefined reference to `xnarch_atomic_set' > > > > collect2: ld returned 1 exit status > > > > What is wrong with this? TSysMain does not exist, as ld removes it after the error. Do I have to specify -r as option to ld?
Best regards -- Niklaus Giger
