On Aug 10, 2009, at 10:11 AM, Lenny Verkhovsky wrote:
> Don't these allocations of bshe->smbhe_keys require some kind of
> memory translation from 1 proc's memory space to another ( in
> bootstrap_init function /ompi/mca/coll/sm/coll_sm_module.c )
> If local rank0 allocates ( get attached to ) memory, others can't
> read it without proper tranlsation.
Ugh -- you may be completely right here. If that's the case, I'm a
total bozo; I don't know how I would have missed that when I wrote
this stuff several years ago. But it would explain some of the
behavior that we have been seeing. I'll dig; thanks.