- 16 Dec, 2014 1 commit
-
-
Xin Zhao authored
Arrange RMA sync functions in src/mpid/ch3/src/ch3u_rma_sync.c in the following order: Win_fence Win_post Win_start Win_complete Win_wait Win_test Win_lock Win_unlock Win_flush Win_flush_local Win_lock_all Win_unlock_all Win_flush_all Win_flush_local_all Win_sync No reviewer.
-
- 11 Dec, 2014 1 commit
-
-
Charles J Archer authored
-
- 09 Dec, 2014 2 commits
-
-
Charles J Archer authored
-
Charles J Archer authored
-
- 08 Dec, 2014 1 commit
-
-
This patch provides the following fix wrt Windows conformance feature (makes single code working on both platforms Linux and Windows): - RMA mutexes fix for Windows Change-Id: Ib4f7b2ec8a07813f0ed35281a1d584637c84c0a9 Signed-off-by:
Ken Raffenetti <raffenet@mcs.anl.gov>
-
- 02 Dec, 2014 1 commit
-
-
Type of the third argument for MPIR_Pack_size_impl should be a pointer to MPI_Aint. This patch fixes the wrong usage of int pointer for MPIR_Pack_size_impl in NEWMAD and MXM netmods. Signed-off-by:
Sangmin Seo <sseo@anl.gov>
-
- 28 Nov, 2014 1 commit
-
-
Pavan Balaji authored
Signed-off-by:
Wesley Bland <wbland@anl.gov>
-
- 26 Nov, 2014 2 commits
-
-
Wesley Bland authored
The function to convert the group of failed procs to a bitarray was incorrectly quiting early if one of the globally known failed processes was not in the communciator being dealt with. Signed-off-by:
Ken Raffenetti <raffenet@mcs.anl.gov>
-
Kenneth Raffenetti authored
We simply use PMI_Abort in both the sock and nemesis code. Remove extra functions and constants that are not useful. Refs #1537 Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
- 24 Nov, 2014 1 commit
-
-
Xin Zhao authored
It is possible that a request handler of RMA request is called for the second time inside the first called request handler on the same request. Consider the following case: a req is queued up in Nemesis SHM queue with ref count of 2: one is for request completion and another is for dequeueing from SHM queue. The first called req handler completed this request and decrement ref count to 1. This request is still in the queue. However, within this handler, we trigger the same req handler on the same request again (for example making progress on SHM queue), and the second called handler also tries to complete this request, which leads to the wrong execution. In this patch we check if request has already been completed when entering the req handler, to prevent processing the same request twice. We also move the function finish_op_on_target() (where the same req handler can be triggered again) after request completion routine, so that we can mark the current request as completed before enter the same req handler for the second time. Fix #2204 Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
- 21 Nov, 2014 1 commit
-
-
* Implements a tag matching interface netmod over the OFIWG Scalable Fabric Interfaces (SFI)
-
- 20 Nov, 2014 1 commit
-
-
Min Si authored
If user does not explicitly set alloc_shm to TRUE in win_create, we should never detect SHM windows because of expensive overhead. However, current code does not check this info flag. This patch fixed it. Closes #2161 Signed-off-by:
Xin Zhao <xinzhao3@illinois.edu>
-
- 19 Nov, 2014 3 commits
-
-
Kenneth Raffenetti authored
Signed-off-by:
Antonio J. Pena <apenya@mcs.anl.gov>
-
Kenneth Raffenetti authored
Signed-off-by:
Antonio J. Pena <apenya@mcs.anl.gov>
-
Kenneth Raffenetti authored
It is possible that PtlMEAppend can return a PTL_NO_SPACE error, meaning there are too many outstanding operations already active. To avoid an abort we simply retry after processing events that have queued up locally. Signed-off-by:
Antonio J. Pena <apenya@mcs.anl.gov>
-
- 17 Nov, 2014 1 commit
-
-
Kenneth Raffenetti authored
-
- 14 Nov, 2014 5 commits
-
-
Antonio Pena Monferrer authored
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>
-
Pavan Balaji authored
Now, when we pop an event, we queue up the buddy event (e.g., ACK for SEND) to return next. This way, we don't need to search for the event everytime. Since we know that there'll be at most one such pending event, we maintain a single event structure for this. Signed-off-by:
Ken Raffenetti <raffenet@mcs.anl.gov>
-
Pavan Balaji authored
We were stashing events when the origin receives a NACK. This is unnecessary since we retransmit the op and never use those stashed events. Signed-off-by:
Ken Raffenetti <raffenet@mcs.anl.gov>
-
Pavan Balaji authored
Signed-off-by:
Ken Raffenetti <raffenet@mcs.anl.gov>
-
Pavan Balaji authored
1. Moved op management to a different file. 2. Move rptl_info to an extern, so it can be shared by multiple files. 3. Separate out rptl initialization routines. Signed-off-by:
Ken Raffenetti <raffenet@mcs.anl.gov>
-
- 13 Nov, 2014 11 commits
-
-
Xin Zhao authored
ReqHandler_GaccumLikeSendComplete is used for GACC-like operations, including GACC, CAS and FOP. Here we split it into following three functions: ReqHandler_GaccumSendComplete ReqHandler_CASSendComplete ReqHandler_FOPSendComplete It is convenient for us to add different actions in future for those three kinds of operations. Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
Xin Zhao authored
Here we wrap up common action when one RMA op is finished on target into a function to make code structure cleaner. Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
Xin Zhao authored
Originally do_accumulate_op() only accepts request pointer as argument which is too restrict to be reused. Here we modify it to access buffer address, count, datatype and op, so that it can be reused in more general cases. Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
Xin Zhao authored
Here we initalize packet flag as FLAG_NONE when creating this packet, and add flags later when needed. Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
Xin Zhao authored
When operation pending list and request lists are all empty, FLUSH message needs to be sent by origin only when origin issued PUT/ACC operations since the last synchronization calls, otherwise origin does not need to issue FLUSH at all and does not need to wait for FLUSH ACK message. Similiarly, origin waits for ACK of UNLOCK message only when origin issued PUT/ACC operations since the last synchronization calls. However, UNLOCK message always needs to be sent out because origin needs to unlock the target process. This patch avoids issuing unnecessary FLUSH / FLUSH ACK / UNLOCK ACK messages. Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
Kenneth Raffenetti authored
Helps clarity since we no longer use ACKs in the netmod code. Signed-off-by:
Antonio Pena Monferrer <apenya@mcs.anl.gov>
-
Antonio Pena Monferrer authored
The rportals layer is taking care of retransmissions, so we should only be interested in delivery events in the netmod layer. Signed-off-by:
Ken Raffenetti <raffenet@mcs.anl.gov>
-
Antonio Pena Monferrer authored
Those were introduced for a robust protocol during development. No longer needed. Signed-off-by:
Ken Raffenetti <raffenet@mcs.anl.gov>
-
Pavan Balaji authored
The user pointer was set, but later overwritten with an internal value. Signed-off-by:
Antonio Pena Monferrer <apenya@mcs.anl.gov>
-
Wesley Bland authored
Depending on how some uninitialized data was prepopulated, the ibsend check was periodically crashing out in the call to MPID_Request_is_pending_failure. Some simple sanity checking to make sure the input data wasn't NULL takes care of this. Signed-off-by:
Huiwei Lu <huiweilu@mcs.anl.gov>
-
Kenneth Raffenetti authored
The previous code only detected a datatype mismatch when the message was copied out of the unexpected queue. Now it will throw an error in both cases. We also set the error in the status object to match the default ch3 behavior. This fixed an issue where the request would not be freed and cause extra debugging output at MPI_Finalize. Signed-off-by:
Antonio Pena Monferrer <apenya@mcs.anl.gov>
-
- 12 Nov, 2014 8 commits
-
-
Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
Pavan Balaji authored
The terminology "flow_control" was a bit of a misnomer since we do more than just enable/disable flow control based on whether messages are on the data or control portal. Signed-off-by:
Antonio Pena Monferrer <apenya@mcs.anl.gov>
-
Pavan Balaji authored
Signed-off-by:
Antonio Pena Monferrer <apenya@mcs.anl.gov>
-
Pavan Balaji authored
We now use a target structure for each target ID that we want to send data to. This allows us to separate out target-specific states and more cleanly manage operations to a single target. Signed-off-by:
Antonio Pena Monferrer <apenya@mcs.anl.gov>
-
Signed-off-by:
Antonio Pena Monferrer <apenya@mcs.anl.gov>
-
Signed-off-by:
Pavan Balaji <balaji@anl.gov>
-
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>
-
Huiwei Lu authored
Similar to d086ac27, check the state of a VC to see if it is valid before creating a group, request or communicator in MPID_Recv. Signed-off-by:
Wesley Bland <wbland@anl.gov>
-