Something screwed up in PROJECT_ROOT or PROJECT_TOPdir

Peter Dufault dufault at hda.com
Sat Feb 19 17:10:42 UTC 2005


On Feb 19, 2005, at 11:31 AM, Ralf Corsepius wrote:

> On Sat, 2005-02-19 at 10:12 -0500, Peter Dufault wrote:
>


> * As you seem to be using rtems-cvs, I strongly recommend you to use
>  maintainer-mode (configure --enable-maintainer-mode).

Will do.

>
> * The CFLAGS you are using are dubious:
> -mcpu=750 -D__ALTIVEC__ -Wa,-m7450
>
> Seems to me as if you are trying to outsmart an older GCC, which does
> not support the 7450, to "fake" the m7500.

OK, which gcc should I use?  Can I look up which versions of the tools 
are in use for which versions of RTEMS?  That would help a lot.  (This 
was gcc 3.4.0 from the RTEMS site).

>
>> Host info, configuration, etc, in the attachment.
> I'd recommend you to start anew (remove the build directory),
> bootstrap the source-tree and configure with --enable-maintainer-mode.
> Also make sure not to have set any PROJECT_* variable in your
> environment or in your <BSP>.cfg.

Will do.

>
>> XXX Compilation failure:
>>
>> if powerpc-rtems-gcc --pipe -B../../../lib/ -B../../../mvme5500/lib/ -
>> specs bsp_specs -qrtems -DPACKAGE_NAME=\"rtems-c-src\" -
>> DPACKAGE_TARNAME=\"rtems-c-src\" -DPACKAGE_VERSION=\"4.6.99.1\" -
>> DPACKAGE_STRING=\"rtems-c-src\ 4.6.99.1\" -DPACKAGE_BUGREPORT=\"rtems-
>> bugs at rtems.com\"  -I.
>> -I../../../../../../tools/rtems-4.7/c/src/optman
>> -isystem ../../../mvme5500/lib/include
>>
>>   -Wall -mcpu=750 -D__ALTIVEC__
>> -Dmpc750 -mmultiple -mstring -mstrict-align -Wa,-m7450 -O4 -MT
>> sapi/no_ext_rel-no-ext.o -MD -MP -MF "sapi/.deps/no_ext_rel-no-
>> ext.Tpo" -c -o sapi/no_ext_rel-no-ext.o `test -f 'sapi/no-ext.c' ||
>> echo '../../../../../../tools/rtems-4.7/c/src/optman/'`sapi/no-ext.c;
>> \
>> then mv -f "sapi/.deps/no_ext_rel-no-ext.Tpo" "sapi/.deps/no_ext_rel-
>> no-ext.Po"; else rm -f "sapi/.deps/no_ext_rel-no-ext.Tpo"; exit 1; fi
>> ../../../../../../tools/rtems-4.7/c/src/optman/sapi/no-ext.c:15:26:
>> rtems/system.h: No such file or directory
>> ../../../../../../tools/rtems-4.7/c/src/optman/sapi/no-ext.c:16:32:
>> rtems/rtems/status.h: No such file or directory
>
> Symptoms of a failed "make preinstall".
>
>> XXX Layout of include files in build directory:
>> XXX Note the two lib/includes.
>>
>> ./mvme5500/lib/include
> Where does this come from? This is not supposed to be there.

I know, it has all the files in it that should be one level lower down.

>
>> ./powerpc-rtems/c/mvme5500/cpukit/score/include
>> ./powerpc-rtems/c/mvme5500/lib/libcpu/powerpc/shared/include
>> ./powerpc-rtems/c/mvme5500/lib/libbsp/powerpc/mvme5500/include
>> ./powerpc-rtems/mvme5500/lib/include
> This one is the only one which is of any importance.

Right, where the files should be.  That's why I thought something with 
lots of ".." in it was messed up.

>
>> /usr/home/dufault/work/rtems_builder/powerpc-rtems/rtems-4.7/powerpc-
>> rtems/c/mvme5500/lib/libcpu/powerpc/shared
>>
>> XXX Configuration
>>
>>   $ ../../tools/rtems-4.7/configure --target=powerpc-rtems --
>> prefix=/opt/rtems-4.7 --enable-posix --enable-networking --enable-cxx
>> --enable-rtems-debug --disable-itron --enable-rtemsbsp=mvme5500
> I can't reproduce your issue.
>
>> FreeBSD
> Wasn't it you who had problems with a broken m4?

Yes, with m4 1.4.2 straight from gnu.org.  All the tools I'm using are 
from the RTEMS website or GNU, and I've had success building things in 
December and last summer (but never MVME5500). If you're willing to 
build gm4 1.4.2 with debugging I've attached a .gdbinit that will tell 
us if the problem is there (sprintf %-349s into 256 bytes on the stack) 
on other hosts but just not causing a core dump.

I'll give up and go to 4.6.2 soon with the Brookhaven patches for 
MVME5500 but I'll try a little more with 4.7.

Peter

Peter Dufault
HD Associates, Inc.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: gm4_gdbinit.txt
URL: <http://lists.rtems.org/pipermail/users/attachments/20050219/73dfb954/attachment-0001.txt>
-------------- next part --------------



More information about the users mailing list