I see target linux-alpha, but no linux-alpha-thr or linux-alpha-thr-ogl. It also seems linux-alpha-lam and linux-alpha-lam-ogl is missing. Can you please check this; Eventually ask upstream!? I can provide testbuilds and/or account to the alpha koji buildsys if needed!
It's trivial to fix. I'll take care of this sometime soon and I'll send an updated patch to upstream after we've tested it.
Great! Thx!
any chance to get this done soon?
Yes, almost done. What does rpm --eval %{_target_cpu} show on Alpha?
[root@gosa ~]# rpm --eval %{_target_cpu} alpha
This should do it then. cvs up ;) Please close when you've tested it.
http://buildsys.zero42.at/koji/getfile?taskID=73766&name=build.log There's still a problem. Some -m64 made it into compilation and alpha gcc doesn't understand that option (well, it better should ignore it, I know, but...) :-)
... mpicc -Wall -O2 -I/usr/lib/lam/h -DLinux -DMPI -DUSEJPEG -DUSEPNG -I/usr/include/libpng12 -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions --param=ssp-buffer-size=4 -mieee -c ../src/api.c -o ../compile/linux-alpha-lam/libtachyon/api.o cc1: error: unrecognized command line option "-m64" mpicc: No such file or directory ... This is a mystery. There's no "-m64" in the command line, so my guess is that mpicc (lam) is adding that on its own. Having no working alpha Linux boxes, I can't help here.
Well. lam isn't very alpha ready, so I do the reassign. :-) Why not ready. Because of that: [oliver@gosa alpha]$ rpm -qpl lam-devel-7.1.2-10.fc8.alpha.rpm |grep conf /usr/include/lam/64/lam_config.h /usr/include/lam/lam_config_bottom.h /usr/lib/pkgconfig/lam.pc If some program (like tachyon) request #include <lam_config.h>, how should gcc pick that up? However, the -m64 that might come from lam as well, needs fixing also. Can you please have a look? Thx.
(In reply to comment #9) > Well. lam isn't very alpha ready, so I do the reassign. :-) Well I don't know about that... > Why not ready. Because of that: > [oliver@gosa alpha]$ rpm -qpl lam-devel-7.1.2-10.fc8.alpha.rpm |grep conf > /usr/include/lam/64/lam_config.h > /usr/include/lam/lam_config_bottom.h > /usr/lib/pkgconfig/lam.pc > > > If some program (like tachyon) request #include <lam_config.h>, how should gcc > pick that up? However, the -m64 that might come from lam as well, needs fixing also. > > Can you please have a look? Thx. The answer is to use pkg-config: $ pkg-config --cflags lam -I/usr/include/lam -I/usr/include/lam/64 And I think there are more problems in tachyon, so I'll reassign it back.
OK, try 0.98-0.5 build from devel.
Well. I think I've found the buggy function from lam: ./lam-7.1.2-rh1/tools/wrappers/wrap.cc: -> lam_wrap_build_cflags does the following: cflags.push_back("-m" + mode); And exactly this - I believe - produces the -m64; I think we should put this into some #ifdef __alpha__.
Ah. It's all in this patch: lam-7.1.2-archinc.patch
Created attachment 289763 [details] Update patch that add's #ifndef __alpha__ Oh yes. The #ifndefs made it!
Patch is - of course - for lam!
Great. I think we should let the lam maintainer know about it, then. Perhaps in a new bug report. May I close this one as fixed?
Yes, please move the changes from rawhide to f8 and close this bug via bodhi afterwards!
tachyon-0.98-0.5.20070319.fc8 has been pushed to the Fedora 8 testing repository. If problems still persist, please make note of it in this bug report. If you want to test the update, you can install it with su -c 'yum --enablerepo=updates-testing update tachyon'
tachyon-0.98-0.5.20070319.fc8 has been pushed to the Fedora 8 stable repository. If problems still persist, please make note of it in this bug report.