GitLab

We were using defines instead of enum to represent the same class of
flags.
Signed-off-by: Wesley Bland <wbland@anl.gov>

We were using an enum for packet types and used an int16_t for the
storage.  Instead we should directly use the enum as the storage.
Signed-off-by: Wesley Bland <wbland@anl.gov>

This is to help with debugging.  Zero is too common a value, and is
often set automatically by the system if not initialized.  Starting at
a different value helps us catch uninitialized cases more easily.

We pick "42" as our magic number as it is the answer to the ultimate
question of life, the Universe, and everything.
Signed-off-by: Wesley Bland <wbland@anl.gov>

Bad placement of commas was making indent very unhappy.
Signed-off-by: Wesley Bland <wbland@anl.gov>

This reverts commit 274a5a70.

We were creating duplicating information in the operation structure
and in the packet structure when the message is actually issued.
Since most of the information is the same anyway, this patch just
embeds a packet structure into the operation structure.
Signed-off-by: Xin Zhao <xinzhao3@illinois.edu>

When the MPI_MODE_NOCHECK assertion is given to a passive target lock
operation, we defer acquisition of the lock and piggyback the request on
the first RMA op to the target.  This eliminates a round-trip
lock-request message.

Reviewer: goodell

Removed source_win_handle from the packet header, since it's no longer
needed.

Reviewer: goodell

This patch uses packet header flags to piggyback the unlock operation on other
RMA operations.  For most operations, there is no net change.  However, FOP and
GACC, unlock piggybacking was previously not implemented.

Reviewer: goodell

Added flags field to RMA operation packets that are sent from origin to target.
This will be used to piggyback RMA synchronization operations.

Reviewer: goodell

Convert the packet type field from an enum to a uint16.  This change was also
applied to packet types defined by Nemesis.  Downstream netmod developers will
also have to make this change if they defined new packet types.

Reviewer: goodell

Reviewer: buntinas

This commit implements MPI-3 RMA's flush and flush_all operations.

Reviewer: buntinas

When enabled, this mode of operation immediately requests the lock when
MPI_Win_lock is called.  Currently, this is enabled by setting the
MPICH_RMA_LOCK_IMMED environment variable.  In the future, we can also make
this mode of operation available though an info/assert.  This capability is
needed to implement MPI-3's flush operations.

Reviewer: buntinas

Reviewer: goodell

By setting "indent-tabs-mode:nil" we should hopefully begin to slowly
squeeze out hard tabs from the source without a disruptive (to
downstream projects) whitespace-fixing change.

No reviewer.

Updated the fetch_and_op implementation to have two data transfer paths; one
where data can be embedded in the packet header and one where it is sent
separately.  With this change, the header size is back to 40 bytes.

Reviewer: buntinas

An extra argument was added to the oputil macros to provide a name,
which can be used to generate union types.  This is needed for
generating types to be included in the RMA packet headers.  The RMA code
is also modified to use this new functionality, replacing the old
hand-coded unions.

Reviewer: goodell@

Adds implementation of atomic fetch-and-op operation and several test cases.

Reviewer: buntinas

The source_win_handle is used to inform the target that the RMA access epoch
has ended.  This was removed from the CAS packet to save space and was replaced
with another packet type, MPIDI_CH3_PKT_CAS_UNLOCK, which conveys the same
information.

This patch adds support for the new MPI-3 compare-and-swap routine.  The
implementation embeds the data into the packet header to reduce the number of
messages.

Reviewer: buntinas

[svn-r7416] Major improvement to RMA performance for long lists of operations, an immediate mode accumulate for single ints, store the MPID_Comm within the window, and added a basic performance instrumentation interface that was extensively used to improve the RMA performance (enabled with --enable-g=instr).  With these fixes, MPICH2 can run the one-sided version of the Graph500 benchmark at a respectable if not great rate