GitLab

MPID_Segment_pack/unpack should be all we need to manipulate noncontig
messages. Not sure why these were originally included.
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

No reviewer.

Min Si authored Jun 10, 2015 and Pavan Balaji committed Jun 14, 2015

This patch includes three changes:
(1) Added netmod API get_ordering to allow netmod to expose the network
ordering. A netmod may issue some packets via multiple connections in
parallel if those packets (such as RMA) do not require ordering, and
thus the packets may be unordered. This patch sets the network ordering
in every existing netmod (tcp|mxm|ofi|portals|llc) to true, since all
packets are sent orderly via one connection.
(2) Nemesis exposes the window packet orderings such as AM flush
ordering at init time. It supports ordered packets only when netmod
supports ordered network.
(3) If AM flush is ordered (flush must be finished after all previous
operations), then CH3 RMA only requests FLUSH ACK on the last operation.
Otherwise, CH3 must request per-OP FLUSH ACK to ensure all operations
are remotely completed.
Signed-off-by: Xin Zhao <xinzhao3@illinois.edu>
Signed-off-by: Pavan Balaji <balaji@anl.gov>

Xin Zhao authored May 14, 2015 and Pavan Balaji committed May 30, 2015

There are two APIs that are used to issue data with a request
created and passed:

iSendv() --- issue contiguous data;
sendNoncontig_fn() --- issue non-contiguous data;

In this patch, we modify the implementation of those two functions
in nemesis and netmod (tcp/mxm/ptl) to make them issue the extended
packet header stored in the request.
Signed-off-by: Pavan Balaji <balaji@anl.gov>

Xin Zhao authored May 13, 2015 and Pavan Balaji committed May 30, 2015

Signed-off-by: Pavan Balaji <balaji@anl.gov>

Two tag bits are reserved for error propagation. We need to make sure
they are ignored by the network matching capabilities. Refs #2260

No reviewer.

Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

The 2 commits being reverted introduced a "safe" PtlMEAppend function
that would call MPID_nem_ptl_poll to process some events in case there
was no space to append the match list entry. However the poll function
is not reentrant safe, which could lead to ordering problems.

The increased list entry limit from [c6c0d6f6

] should prevent PTL_NO_SPACE
errors from happening, except in the extreme case. If we still find we are
hitting this error, a proper fix can be done in the Rportals layer.
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

Going from a macro to a function fixes the issue because of creating a
copy of the pointer.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Helps clarity since we no longer use ACKs in the netmod code.
Signed-off-by: Antonio Pena Monferrer <apenya@mcs.anl.gov>

Kenneth Raffenetti authored Nov 11, 2014 and Pavan Balaji committed Nov 12, 2014

Signed-off-by: Antonio Pena Monferrer <apenya@mcs.anl.gov>

Antonio Pena Monferrer authored Nov 11, 2014 and Pavan Balaji committed Nov 12, 2014

Signed-off-by: Pavan Balaji <balaji@anl.gov>

Antonio J. Pena authored Oct 22, 2014 and Kenneth Raffenetti committed Nov 12, 2014

Full redesign, mainly of the functions in ptl_nm.c and the
communications involving the "control" portal. Still some
problems with flow control.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

All MPI_Sends in the Portals4 netmod will cause some or all of the data to be
sent eagerly to the receiver. Canceling a send means having to find the data in
the unexpected message queue and removing it in order to preserve matching.
Because the message queues exist at the netmod level, it needs its own cancel
protocol.

The protocol is modeled on a similar case in CH3, but with its own method
for searching the unexpected queue. Custom netmod packet handlers are used to
receive and process the control messages.

Known Issue:
  Because we are using different PTs for the send and cancel message, it is
  possible the cancel request could arrive before the message being canceled.
Signed-off-by: Antonio Pena Monferrer <apenya@mcs.anl.gov>

Kenneth Raffenetti authored Nov 04, 2014 and Pavan Balaji committed Nov 05, 2014

The large send handler incorrectly assumed event ordering from portals.
This could lead to a request being freed while pending events would
still attempt to access it, causing a segfault or incorrect handler to
execute.
Signed-off-by: Pavan Balaji <balaji@anl.gov>

Kenneth Raffenetti authored Oct 28, 2014 and Pavan Balaji committed Nov 05, 2014

In large message cases, when multiple get operations are issued, the
data may arrive out-of-order back at the initiator. A counter is required
to ensure all operations have completed. In the temporary buffer case, we
simply wait for all the data to arrive, and unpack in one operation.
Signed-off-by: Pavan Balaji <balaji@anl.gov>

Kenneth Raffenetti authored Oct 23, 2014 and Pavan Balaji committed Nov 05, 2014

Large messages (either larger than max_msg_size or > MPID_IOV_LIMIT), will
be packed into a temporary buffer. These need to be optimized.
Signed-off-by: Pavan Balaji <balaji@anl.gov>

Kenneth Raffenetti authored Oct 10, 2014 and Pavan Balaji committed Nov 05, 2014

If a message is larger than the max_msg_size limit, issue multiple MEs
for the remainder of the message. Completion events for the intermediate
operations will be ignored. Only the final operation will trigger the
event handler to tell MPI that communication is complete.
Signed-off-by: Pavan Balaji <balaji@anl.gov>

An EQ for origin events is useful for rate-limiting operations so that
a process does not locally trigger a flow control event on its portal.
We will implement the rate-limiting logic in the rportals layer.
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

Pavan Balaji authored Oct 21, 2014 and Kenneth Raffenetti committed Nov 03, 2014

Portals4 by itself does not provide any flow-control.  This needs to
be managed by an upper-layer, such as MPICH.  Before this patch we
were relying on a bunch of unexpected buffers that were posted to the
portals library to manage unexpected messages.  However, since portals
asynchronously pulls out messages from the network, if the application
is delayed, it might result in the unexpected buffers being filled out
and the portal disabled.  This would cause MPICH to abort.

In this patch, we implement an initial version of flow-control that
allows us to reenable the portal when it gets disabled.  All this is
done in the context of the "rportals" wrappers that are implemented in
the rptl.* files.  We create an extra control portal that is only used
by rportals.  When the primary data portal gets disabled, the target
sends PAUSE messages to all other processes.  Once each process
confirms that it has no outstanding packets on the wire (i.e., all
packets have either been ACKed or NACKed), it sends a PAUSE-ACK
message.  When the target receives PAUSE-ACK messages from all
processes (thus confirming that the network traffic to itself has been
quiesced), it reenables the portal and sends an UNPAUSE message to all
processes.

This patch still does not deal with origin-side resource exhaustion.
This can happen, for example, if we run out of space on the event
queue on the origin side.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

PtlNIInit can optionally return the limitations of a network interface.
Get these limits so we can account for things like the max_msg_size.
Signed-off-by: Antonio Pena Monferrer <apenya@mcs.anl.gov>

First working version. See #2152.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

The out-of-interest bits must be zeroed-out to avoid them colliding with their neighbor bits.
This is relevant In cases of special values, i.e., negative values such as MPI_*_ANY.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Based on an Intel contributed patch.  The idea is to use the bits from
the cancelled field to extend the count, rather than increasing the
count datatype itself.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Fixes to the bit manipulation based on feedback from Artem Yalozo @
Intel.

Fixes to the naming convention based on feedback from Bill Gropp.
Signed-off-by: William Gropp <wgropp@illinois.edu>

[svn-r10770] removed channel_private field in VC and used MPIDI_CH3_VC_DECL macro which is overridden by channel. Reviewed by Dinan

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.