goodell

commit f35d474f0cc42d31f5f4ffb50261362a6747ca98
Author: Dave Goodell <[email protected]>
Date:   Wed Jan 18 18:26:22 2012 -0600

    --enables-strict=posix improvements: choose standard & make 2001 default
    
    You can now choose from three new "--enable-strict=" values:
    "posix1995", "posix2001", and "posix2008".
    
    Also the default has been changed from 1995 to 2001.  This resolves a number

goodell

commit 292bab7c2aa308bd00041374551090f7c0d30746
Author: Dave Goodell <[email protected]>
Date:   Tue Feb 14 14:35:00 2012 -0600

    implement MPIX_Mprobe and friends
    
    This includes MPIX_Mprobe, MPIX_Improbe, MPIX_Mrecv, and MPIX_Imrecv.
    
    !!!NOTE!!! This change adds three more functions to the ADI.
    

goodell

commit bd18e1cf379cea00ba56f87b36b5b869b93d99bb
Author: Dave Goodell <[email protected]>
Date:   Wed Mar 7 16:23:45 2012 -0600

    strengthen mprobe test checks
    
    No reviewer.
---
 test/mpi/pt2pt/mprobe.c |    8 ++++++++
 1 file changed, 8 insertions(+)

goodell

commit 7f49a32b72ff868d46da29fe345fe5c5c0c0ca83
Author: Dave Goodell <[email protected]>
Date:   Wed Mar 7 16:18:50 2012 -0600

    add a helper macro to create "proc null" requests
    
    Convert to using it in CH3's MPID_Irecv
    
    No reviewer.
---

goodell

commit 62568dc4391e004f7d648faeca1ee543fb5561e4
Author: Dave Goodell <[email protected]>
Date:   Wed Mar 7 16:23:19 2012 -0600

    mprobe: support MPI_PROC_NULL/MPIX_MESSAGE_NO_PROC
    
    This corresponds to the functionality described in MPI Forum ticket #328.
    
    No reviewer.
---

goodell

commit 4eb0e178cc83589389f2faaa3a2526d222c155f0
Author: Dave Goodell <[email protected]>
Date:   Mon Mar 12 11:24:35 2012 -0500

    fix mpich1 'errhand' test broken by r9588
    
    I oversimplified the error handling code for MPI_Comm_create and
    MPIX_Comm_create_group that went into that "merge" commit.
    
    Reviewed by dinan@.

goodell

commit e97e6041b9f5fb623b9357f1ff0f73c1b50fb1ef
Author: Dave Goodell <[email protected]>
Date:   Thu Apr 12 13:15:03 2012 -0500

    tt#1599: eliminate MPIDI_CH3_PktGeneric_t
    
    Without this change we are vulnerable to a nasty bit of compiler
    behavior.  MPIDI_CH3_PktGeneric_t sometimes included padding between the
    "kind" and "pktptrs" fields.  Then whole-structure assignments using the
    generic packet (e.g., ch3_istartmsg.c:96) of buffers that are actually

goodell

commit 7592f7d3de4b8da5a33d387e74d87753608d512c
Author: Dave Goodell <[email protected]>
Date:   Thu Apr 12 13:15:03 2012 -0500

    tt#1599: eliminate MPIDI_CH3_PktGeneric_t
    
    Without this change we are vulnerable to a nasty bit of compiler
    behavior.  MPIDI_CH3_PktGeneric_t sometimes included padding between the
    "kind" and "pktptrs" fields.  Then whole-structure assignments using the
    generic packet (e.g., ch3_istartmsg.c:96) of buffers that are actually

goodell

commit a5566493e8acfca25718cbf6539c562c52ddbf3e
Author: Dave Goodell <[email protected]>
Date:   Thu Apr 19 18:07:54 2012 -0500

    use better algorithms in MPIR_Allreduce_group
    
    This is a copy of the non-SMP-aware code in MPIR_Allreduce_intra that
    has been modified to perform rank translation immediately prior to any
    communication calls.  For cases where incremental group rank translation
    is abnormally slow compared to batch translation, this is a non-optimal

goodell

static int
write_NC(NC *ncp)
{   
    int status, mpireturn, rank;
    void *buf; 
    MPI_Offset hsz; /* header size with 0-padding if needed */
    MPI_Status mpistatus;
 
    assert(!NC_readonly(ncp));