[O-MPI devel] ltdl.h problem
As I continue to have the same problem with the missing ltdl.h header I reported few days ago, I spend some time today to dig a little bit inside to find out what and how happens. Finally, I figure out the problem. It happens only after the last set of changes on the Makefile.am and only on systems where there is not a ltdl.h installed on a default location. Otherwise ... on systems where the ltdl.h is installed ... the ugliest things may happens. We can use our ltdl.h in some directories (like the opal base) and the system one in others (because a missing -I). How ? Simply because most of the base components require the opal/mca/mca_base_component_repository.h include. In this file at line 27 there is a include "ltdl.h" supposely protected by the OMPI_WANT_LIBLTDL. This define is always true as it come from the ompi_config.h so the ltdl.h is always required. Now on system where this file does not exist on some system include directories we are supposed to get the one we have ... except that the -I for our include is not added in all the base directories after the last set of changes to the Makefile.am. Now your turn to have fun ... and to bring back a consistent behaviour :) Thanks, george. "We must accept finite disappointment, but we must never lose infinite hope." Martin Luther King
Re: [O-MPI devel] ltdl.h problem
Hi George, * George Bosilca wrote on Wed, Nov 23, 2005 at 08:15:30AM CET: > As I continue to have the same problem with the missing ltdl.h header I > reported few days ago, I spend some time today to dig a little bit inside > to find out what and how happens. Finally, I figure out the problem. It > happens only after the last set of changes on the Makefile.am and only on > systems where there is not a ltdl.h installed on a default location. > Otherwise ... on systems where the ltdl.h is installed ... the ugliest > things may happens. We can use our ltdl.h in some directories (like the > opal base) and the system one in others (because a missing -I). How ? > Simply because most of the base components require the > opal/mca/mca_base_component_repository.h include. In this file at line 27 > there is a include "ltdl.h" supposely protected by the OMPI_WANT_LIBLTDL. > This define is always true as it come from the ompi_config.h so the ltdl.h > is always required. Now on system where this file does not exist on some > system include directories we are supposed to get the one we have ... > except that the -I for our include is not added in all the base > directories after the last set of changes to the Makefile.am. Hmm. Wasn't the decision a while ago to #include consistently, plus, in order to allow the next version of libltdl to work seamlessly as well, to also -I.../libltdl (although the Libtool documentation suggests otherwise)? I haven't followed changes regarding this closely, but above would be safe for OpenMPI in both cases: both failure to include the in-tree ltdl.h as well as failure with Libtool-2.0 will result in compilation errors, and are thus easy to find and fix. Cheers, Ralf
[O-MPI devel] 1.0.1rc3 now available
There was a minor problem with some fortran fixes (the F_STATUSES
stuff) and Edgar found a few type mismatches in Fortran API bindings.
rc3 is now up, containing fixes for all of this:
http://www.open-mpi.org/software/v1.0/downloads/openmpi-1.0.1rc3.tar.gz
http://www.open-mpi.org/software/v1.0/downloads/openmpi-1.0.1rc3.tar.bz2
On Nov 22, 2005, at 12:19 PM, Jeff Squyres wrote:
Anthony Chan found another issue this morning that was worth fixing
for 1.0.1 -- this is now rolled into rc2:
http://www.open-mpi.org/software/v1.0/downloads/openmpi-1.0.1rc2.tar.gz
http://www.open-mpi.org/software/v1.0/downloads/openmpi
-1.0.1rc2.tar.bz2
--
{+} Jeff Squyres
{+} The Open MPI Project
{+} http://www.open-mpi.org/
Re: [O-MPI devel] ltdl.h problem
Fixed in SVN. Sorry about that... On Nov 23, 2005, at 2:15 AM, George Bosilca wrote: As I continue to have the same problem with the missing ltdl.h header I reported few days ago, I spend some time today to dig a little bit inside to find out what and how happens. Finally, I figure out the problem. It happens only after the last set of changes on the Makefile.am and only on systems where there is not a ltdl.h installed on a default location. Otherwise ... on systems where the ltdl.h is installed ... the ugliest things may happens. We can use our ltdl.h in some directories (like the opal base) and the system one in others (because a missing -I). How ? Simply because most of the base components require the opal/mca/mca_base_component_repository.h include. In this file at line 27 there is a include "ltdl.h" supposely protected by the OMPI_WANT_LIBLTDL. This define is always true as it come from the ompi_config.h so the ltdl.h is always required. Now on system where this file does not exist on some system include directories we are supposed to get the one we have ... except that the -I for our include is not added in all the base directories after the last set of changes to the Makefile.am. Now your turn to have fun ... and to bring back a consistent behaviour :) Thanks, george. "We must accept finite disappointment, but we must never lose infinite hope." Martin Luther King ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel
Re: [O-MPI devel] ltdl.h problem
The problem seems to be fixed now. Thanks, george. On Nov 23, 2005, at 2:39 PM, Brian Barrett wrote: Fixed in SVN. Sorry about that... On Nov 23, 2005, at 2:15 AM, George Bosilca wrote: As I continue to have the same problem with the missing ltdl.h header I reported few days ago, I spend some time today to dig a little bit inside to find out what and how happens. Finally, I figure out the problem. It happens only after the last set of changes on the Makefile.am and only on systems where there is not a ltdl.h installed on a default location. Otherwise ... on systems where the ltdl.h is installed ... the ugliest things may happens. We can use our ltdl.h in some directories (like the opal base) and the system one in others (because a missing -I). How ? Simply because most of the base components require the opal/mca/mca_base_component_repository.h include. In this file at line 27 there is a include "ltdl.h" supposely protected by the OMPI_WANT_LIBLTDL. This define is always true as it come from the ompi_config.h so the ltdl.h is always required. Now on system where this file does not exist on some system include directories we are supposed to get the one we have ... except that the -I for our include is not added in all the base directories after the last set of changes to the Makefile.am. Now your turn to have fun ... and to bring back a consistent behaviour :) Thanks, george. "We must accept finite disappointment, but we must never lose infinite hope." Martin Luther King ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel ___ devel mailing list de...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/devel "Half of what I say is meaningless; but I say it so that the other half may reach you" Kahlil Gibran
[O-MPI devel] MPI_Probe_tag_c mvapi hand
I'm running the intel test suite against ompi revision r8247 (v1.0 branch), and the MPI_Probe_tag_c test is hanging on IU's thor cluster. This only happens with using mvapi, and not with gm or tcp. The hang happens whether or not I use sm with mvapi. The processes appear to be spinning on the CPU, and a backtrace of one of them looks like the following: (gdb) bt #0 0x40341754 in ioctl () from /lib/libc.so.6 #1 0x404bbe99 in vip_ioctl_wrapper (ops=VIPKL_OPEN_HCA, pi=0x0, pi_sz=0, po=0x0, po_sz=0) at vipkl_sys_user.c:54 #2 0x404bb886 in VIPKL_EQ_poll (usr_ctx=0x0, hca_hndl=0, vipkl_eq=0, eqe_p=0x40de3eb4) at vipkl_wrap_user.c:1676 #3 0x404bc0e1 in eq_poll_thread (eq_pollt_ptr=0x81377f8) at hobul.c:320 #4 0x4024aef6 in pthread_start_thread () from /lib/libpthread.so.0 #5 0x4034823a in clone () from /lib/libc.so.6 I'm not sure this is useful - can someone else reproduce this? If more information is needed, let me know. Andrew
