Open MPI logo

Hardware Locality Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Hardware Locality Development mailing list

Subject: Re: [hwloc-devel] hwloc-1.3.1 assertion failures on Linux/POWER7
From: Paul H. Hargrove (PHHargrove_at_[hidden])
Date: 2012-02-03 13:28:25


On 2/3/2012 5:27 AM, Brice Goglin wrote:
> Is "vendor=ibm" equivalent to "compiler=xlc"?
> The current vendor detection code looks for __xlC__ __IBMC__ and
> __IBMCPP__ before falling back to __AIX without __GNUC__. Can we pass
> -qhalt=e in all these cases? Or should we grep for xlc in the version
> output like [1] ?
>
> Brice

The observation that (__AIX && !__GNU__) -> vendor=ibm sounds like a
safety net for unknown compilers.
I think that since the current problem is specific to the Linux
sched_setaffinity() probe, that ignoring that case is safe.

BTW: output of xlc when given no arguments is an entire manpage, not
just the version info!

It occurs to me now that xlc MIGHT have an option to make only the
specific (E) become fatal.
I will look and report back.

-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