Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] sm latency putback
From: Tim Mattox (timattox_at_[hidden])
Date: 2009-02-17 15:52:43


Hi Eugene,

On Tue, Feb 17, 2009 at 11:03 AM, Eugene Loh <Eugene.Loh_at_[hidden]> wrote:
> I think I just did my first putback to the trunk. God help us all!

Glad to have you helping out!

> It's r20578 and feedback (e.g., "you broke everything") is appreciated,
> gentle feedback even more so.

This commit looks good to me, but I haven't run/built it yet.
Having looked at it some (mostly to see how you did things),
I ran across one small bit of unprotected debugging
code in btl_sm_component.c that should be changed like this:

+ #if OMPI_ENABLE_DEBUG
                if ( FIFO_MAP(peer_smp_rank) != j )
+ {
                    opal_output(0, "mca_btl_sm_component_progress: "
                        "rank %d got %d on FIFO %d, but this sender
should send to FIFO %d\n",
                        my_smp_rank, peer_smp_rank, j, FIFO_MAP(peer_smp_rank));
+ }
+ #endif

That would avoid adding an "if" to the critical code path on receiving
a sm message.

Jeff graciously just added a Developer Configure Options wiki page
that describes OMPI_ENABLE_DEBUG among other useful things:
 https://svn.open-mpi.org/trac/ompi/wiki/developer/ConfigureOptions

Have a look.

-- 
Tim Mattox, Ph.D. - http://homepage.mac.com/tmattox/
 tmattox_at_[hidden] || timattox_at_[hidden]
    I'm a bright... http://www.the-brights.net/