Hi,
This this the log I got when building the GNU tool chain on ubuntu10.4 64-bit OS.
First I checked out all the tool chain stuff from SVN. Initially OR1KSIM was built and installed properly(/opt/OR1KSIM/bin). When I am building tool chain for newlib, it is giving following error. Though I follow the steps given on the GNU tool chain page, it is not building the tool chain. Please look into the log and point out where it is wrong.

../../.././gdb-7.2/sim/or32/wrapper.c: In function �sim_open�:
../../.././gdb-7.2/sim/or32/wrapper.c:175: warning: passing argument 1 of �or1ksim_init� makes pointer from integer without a cast
/opt/or1ksim/include/or1ksim.h:49: note: expected �const char *� but argument is of type �int�
../../.././gdb-7.2/sim/or32/wrapper.c:175: warning: passing argument 2 of �or1ksim_init� from incompatible pointer type
/opt/or1ksim/include/or1ksim.h:49: note: expected �const char *� but argument is of type �char **�
../../.././gdb-7.2/sim/or32/wrapper.c:189: warning: cast to pointer from integer of different size
../../.././gdb-7.2/sim/or32/wrapper.c: In function �sim_resume�:
../../.././gdb-7.2/sim/or32/wrapper.c:650: error: �OR1KSIM_RC_HALTED� undeclared (first use in this function)
../../.././gdb-7.2/sim/or32/wrapper.c:650: error: (Each undeclared identifier is reported only once
../../.././gdb-7.2/sim/or32/wrapper.c:650: error: for each function it appears in.)
make[2]: *** [wrapper.o] Error 1
make[2]: *** Waiting for unfinished jobs....
make[2]: Leaving directory `/home/avamsi/OpenRISC/gnu-src/bd-elf-gdb/sim/or32'
make[1]: *** [all] Error 1
make[1]: Leaving directory `/home/avamsi/OpenRISC/gnu-src/bd-elf-gdb/sim'
make: *** [all-sim] Error 2
make (all-build all-sim all-gdb) failed.


Thanks,
A,Vamsi

Did you pass this or1ksim install directory with --or1ksim-dir to bld-all.sh?

Hi,

I followed the steps from the following page. I did not build it from the bld-all.sh.

http://opencores.org/openrisc,or1ksim#downloading

Can you correct me if my procedure is wrong.

Thanks,
A.vamsi

Hi A Vasami

I tried this on a 64-bit machine running Fedora 16 and it worked for me. These are the versions:

$ uname -a
Linux mauli 3.3.0-4.fc16.x86_64 #1 SMP Tue Mar 20 18:05:40 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux
$ gcc --version
gcc (GCC) 4.6.3 20120306 (Red Hat 4.6.3-2)

I downloaded the latest versions of Or1ksim and the GNU tools from SVN and uClibc and Linux from git. Here is how I built Or1ksim from within the or1ksim directory.

mkdir bd
cd bd
../configure --target=or32-elf --prefix=/opt/or1ksim-new
make
make install

Here is how I built the tool chains from the GNU tools directory. I had downloaded uClibc and Linux into the same directory.

./bld-all.sh --force --prefix /opt/openrisc --or1ksim-dir /opt/or1ksim-new --uclibc-dir uClibc --linux-dir linux

The tools appear to have built correctly

$ PATH=/opt/openrisc/bin:$PATH
$ or32-elf-gcc --version
or32-elf-gcc (OpenRISC 32-bit toolchain for or32-elf (built 20120424)) 4.5.1-or32-1.0rc4

I can only think you do not have the latest source from SVN. If it is still not working, would you post the results of uname -a and gcc --version here.

HTH

Jeremy

--
Tel: +44 (1590) 610184
Cell: +44 (7970) 676050
SkypeID: jeremybennett
Email: jeremy.bennett@embecosm.com
Web: www.embecosm.com

hi Jeremy,

Thank you for your replay,

I found where the problem is coming. Actually, earlier I installed the or1ksim in the /opt/or1ksim. That path was not being taken when I tried to build GNU newlib tool chain. Now I installed the or1ksim somewhere in my home directory. I provided this path. GNU tool chain got built. Now I am able to run some bare-metal apps on simulator.

Earlier I faced two issues.
1) The binaries(or1ksim and or32elf) which are available for download, will not run directly on my machine. I had to checkout all the tool chain source code and build here itself.
2) Though we install or1ksim in /opt/or1ksim, this path is not being taken though we provide it to bld-all.sh when building.

Thanks,
A.vamsi

Hi A Vamsi,

Glad you have solved the problem.

The first issue you raise is a generic problem for binaries. It is why it is always best to compile from source.

Your second issue is a bug. Would you file it in Bugzilla, so it does not get forgotten (http://bugzilla.opencores.org/). If you could also update the Wiki build instructions to note that you should add Or1ksim to the PATH before building the tool chain, that will be helpful.

Best wishes,

Jeremy