Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |  

This web mail archive is frozen.

This page is part of a frozen web archive of this mailing list.

You can still navigate around this archive, but know that no new mails have been added to it since July of 2016.

Click here to be taken to the new web archives of this list; it includes all the mails that are in this frozen archive plus all new mails that have been sent to the list since it was migrated to the new archives.

Subject: Re: [OMPI devel] 1.5.5rc1 tested: MacOS 10.4 x86-64 hwloc build failure
From: Paul H. Hargrove (PHHargrove_at_[hidden])
Date: 2011-12-14 03:04:27


On 12/13/2011 11:50 PM, Brice Goglin wrote:
> Le 14/12/2011 08:29, Paul H. Hargrove a écrit :
>> I've attempted the build on MacOS 10.4 (Tiger) on x86-64, I hit the
>> same hwloc issue I've encountered on {Free,Open,Net}BSD.
>> The build fails with
>>> CCLD opal_wrapper
>>> /usr/bin/ld: Undefined symbols:
>>> _opal_hwloc122_hwloc_backend_sysfs_exit
>>> _opal_hwloc122_hwloc_backend_sysfs_init
>>> _opal_hwloc122_hwloc_look_linux
>>> _opal_hwloc122_hwloc_set_linux_hooks
> At least FreeBSD and darwin should be detected by configure so the big
> configure warning should not appear there.
>
> OpenBSD and NetBSD are not detected explicitly IIRC, so the warning is
> expected there.
>
> But none of these cases should cause the Linux code to be built anyway
> (the above symbols), so it seems to me that something is broken in the
> way OMPI uses hwloc's configure.
>
> Brice
>
>

I appear to have been mistaken about the big configure warning on the
FreeBSD systems.
It was only on OpenBSD and NetBSD.

I've just build a stand-alone hwloc-1.2.2 on FreeBSD-8.2 and it can
"make all install check" w/o any problems.
It even appears to get the right topology:
> *** Printing overall tree
> Machine#0
> Socket#0
> L2#0(4096KB)
> L1#0(32KB)
> Core#0
> PU#0
> Socket#1
> L2#1(4096KB)
> L1#1(32KB)
> Core#0
> PU#1

I can also build hwloc-1.2.2 on NetBSD-5.1 and OpenBSD-5.0, but don't
get the full topology:
> *** Printing overall tree
> Machine#0
> PU#0
> PU#1

So, I agree w/ Brice that OMPI is getting something wrong, as evidenced
by the failures to link linux-specific hwloc symbols.
My reports (other than the one for RHL8) have been unfairly pointing at
hwloc.

-Paul

-- 
Paul H. Hargrove                          PHHargrove_at_[hidden]
Future Technologies Group
HPC Research Department                   Tel: +1-510-495-2352
Lawrence Berkeley National Laboratory     Fax: +1-510-486-6900