The data endpoint was not being set correctly for local peers in some
cases. This commit fixes the bug and cleans the associated code to
simplify the logic.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit moves the info subscribe for the blocking_fence to after
the global_state is allocated and moves setting win->w_osc_module to
before the info subscribe for alloc_shared_contig. This fixes a SEGV
caught by MTT.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
a buffer defined by (buf, count, dt)
will have data starting at buf+offset and ending len bytes later with
len = opal_datatype_span(&dt.super, count, &offset);
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
origin_datatype and target_datatype might be different and hence have different extent,
so use either origin_extent or target_extent when appropriate.
Refs open-mpi/ompi#3569
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
The osc_rdma_get_remote_segment() has the 3rd and 4th args as
* target_disp
* length
which it uses to determine if the rdma falls within the bounds of
the window or not (actually it only checks the upper bound, but I'm
okay with that).
Anyway the caller previously was passing in the length argument as
target_datatype->super.size * target_count
which which doesn't really represent the number of bytes after target_disp
for which data exists. In particular I could create a datatype as
{ disp -4, len 4 } and use target_disp 4
and that would be bytes 0-3 of the window where the original code
would think it was bytes 4-7 and could abort at the range check.
Ive changed it to use the opal_datatype_span() function.
Signed-off-by: Mark Allen <markalle@us.ibm.com>
See bug report
https://github.com/open-mpi/ompi/issues/3548
If a 1sided test is launched -host hostA:2,hostB:1 some of the ranks
call allocate_state_single() and others call allocate_state_shared().
These functions were producing different values for module->state_size
but that's used when they lookup peer info from each other in
ompi_osc_rdma_peer_setup() so they need to all have matching
module->state_offset values.
This change adds a few unused bytes in the memory allocate_state_single()
creates so it matches.
Signed-off-by: Mark Allen <markalle@us.ibm.com>
The expected sequence of events for processing info during object creation
is that if there's an incoming info arg, it is opal_info_dup()ed into the obj
at obj->s_info first. Then interested components register callbacks for
keys they want to know about using opal_infosubscribe_infosubscribe().
Inside info_subscribe_subscribe() the specified callback() is called with
whatever matching k/v is in the object's info, or with the default. The
return string from the callback goes into the new k/v stored in info, and
the input k/v is saved as __IN_<key>/<val>. It's saved the same way
whether the input came from info or whether it was a default. A null return
from the callback indicates an ignored key/val, and no k/v is stored for
it, but an __IN_<key>/<val> is still kept so we still have access to the
original.
At MPI_*_set_info() time, opal_infosubscribe_change_info() is used. That
function calls the registered callbacks for each item in the provided info.
If the callback returns non-null, the info is updated with that k/v, or if
the callback returns null, that key is deleted from info. An __IN_<key>/<val>
is saved either way, and overwrites any previously saved value.
When MPI_*_get_info() is called, opal_info_dup_mpistandard() is used, which
allows relatively easy changes in interpretation of the standard, by looking
at both the <key>/<val> and __IN_<key>/<val> in info. Right now it does
1. includes system extras, eg k/v defaults not expliclty set by the user
2. omits ignored keys
3. shows input values, not callback modifications, eg not the internal values
Currently the callbacks are doing things like
return some_condition ? "true" : "false"
that is, returning static strings that are not to be freed. If the return
strings start becoming more dynamic in the future I don't see how unallocated
strings could support that, so I'd propose a change for the future that
the callback()s registered with info_subscribe_subscribe() do a strdup on
their return, and we change the callers of callback() to free the strings
it returns (there are only two callers).
Rough outline of the smaller changes spread over the less central files:
comm.c
initialize comm->super.s_info to NULL
copy into comm->super.s_info in comm creation calls that provide info
OBJ_RELEASE comm->super.s_info at free time
comm_init.c
initialize comm->super.s_info to NULL
file.c
copy into file->super.s_info if file creation provides info
OBJ_RELEASE file->super.s_info at free time
win.c
copy into win->super.s_info if win creation provides info
OBJ_RELEASE win->super.s_info at free time
comm_get_info.c
file_get_info.c
win_get_info.c
change_info() if there's no info attached (shouldn't happen if callbacks
are registered)
copy the info for the user
The other category of change is generally addressing compiler warnings where
ompi_info_t and opal_info_t were being used a little too interchangably. An
ompi_info_t* contains an opal_info_t*, at &(ompi_info->super)
Also this commit updates the copyrights.
Signed-off-by: Mark Allen <markalle@us.ibm.com>
ompi_communicator_t, ompi_win_t, ompi_file_t all have a super class of type opal_infosubscriber_t instead of a base/super type of opal_object_t (in previous code comm used c_base, but file used super). It may be a bit bold to say that being a subscriber of MPI_Info is the foundational piece that ties these three things together, but if you object, then I would prefer to turn infosubscriber into a more general name that encompasses other common features rather than create a different super class. The key here is that we want to be able to pass comm, win and file objects as if they were opal_infosubscriber_t, so that one routine can heandle all 3 types of objects being passed to it.
MPI_INFO_NULL is still an ompi_predefined_info_t type since an MPI_Info is part of ompi but the internal details of the underlying information concept is part of opal.
An ompi_info_t type still exists for exposure to the user, but it is simply a wrapper for the opal object.
Routines such as ompi_info_dup, etc have all been moved to opal_info_dup and related to the opal directory.
Fortran to C translation tables are only used for MPI_Info that is exposed to the application and are therefore part of the ompi_info_t and not the opal_info_t
The data structure changes are primarily in the following files:
communicator/communicator.h
ompi/info/info.h
ompi/win/win.h
ompi/file/file.h
The following new files were created:
opal/util/info.h
opal/util/info.c
opal/util/info_subscriber.h
opal/util/info_subscriber.c
This infosubscriber concept is that communicators, files and windows can have subscribers that subscribe to any changes in the info associated with the comm/file/window. When xxx_set_info is called, the new info is presented to each subscriber who can modify the info in any way they want. The new value is presented to the next subscriber and so on until all subscribers have had a chance to modify the value. Therefore, the order of subscribers can make a difference but we hope that there is generally only one subscriber that cares or modifies any given key/value pair. The final info is then stored and returned by a call to xxx_get_info.
The new model can be seen in the following files:
ompi/mpi/c/comm_get_info.c
ompi/mpi/c/comm_set_info.c
ompi/mpi/c/file_get_info.c
ompi/mpi/c/file_set_info.c
ompi/mpi/c/win_get_info.c
ompi/mpi/c/win_set_info.c
The current subscribers where changed as follows:
mca/io/ompio/io_ompio_file_open.c
mca/io/ompio/io_ompio_module.c
mca/osc/rmda/osc_rdma_component.c (This one actually subscribes to "no_locks")
mca/osc/sm/osc_sm_component.c (This one actually subscribes to "blocking_fence" and "alloc_shared_contig")
Signed-off-by: Mark Allen <markalle@us.ibm.com>
Conflicts:
AUTHORS
ompi/communicator/comm.c
ompi/debuggers/ompi_mpihandles_dll.c
ompi/file/file.c
ompi/file/file.h
ompi/info/info.c
ompi/mca/io/ompio/io_ompio.h
ompi/mca/io/ompio/io_ompio_file_open.c
ompi/mca/io/ompio/io_ompio_file_set_view.c
ompi/mca/osc/pt2pt/osc_pt2pt.h
ompi/mca/sharedfp/addproc/sharedfp_addproc.h
ompi/mca/sharedfp/addproc/sharedfp_addproc_file_open.c
ompi/mca/topo/treematch/topo_treematch_dist_graph_create.c
ompi/mpi/c/lookup_name.c
ompi/mpi/c/publish_name.c
ompi/mpi/c/unpublish_name.c
opal/mca/mpool/base/mpool_base_alloc.c
opal/util/Makefile.am
since Open MPI now requires a C99, and ptrdiff_t type is part of C99,
there is no more need for the abstract OPAL_PTRDIFF_TYPE type.
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
One should use the correct module object when calling
c_coll.coll_allgather. Otherwise there will be a segfault in the
case, for example, when hcoll is used. In that case
c_coll.coll_allgather = mca_coll_hcoll_allgather while
c_coll.coll_gather_module = tuned.
Signed-off-by: Valentin Petrov <valentinp@mellanox.com>
As we changed the ABI (forcing a major release), we can limit
the size of the predefined communicators by moving the collective
structure outside the communicator. This might have a minimal,
but unnoticeable, impact on performance. This approach has been
discussed during the January 2017 devel meeting.
Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
Signed-off-by: Joshua Hursey <jhursey@us.ibm.com>
Under heavy load the locking code could fail if the underlying btl
module started to return OPAL_ERR_OUT_OF_RESOURCE on atomic
operations. This commit updates the code to gracefully handle btl
errors.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
in this context, AMD64 really means amd64 or em64t, so let's
rename this into X86_64 in order to avoid any confusion
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
This commit implements onesided operations for noncontiguous
datatypes using two different algorithms.
* If the result and/or origin datatype is noncontiguous and the
target datatype is contiguous, then an iovec MD is created for
the result and origin. The operation is performed using a
single Portals4 call (unless it exceeds the max message size).
* If the target datatype is noncontigous, then an algorithm
similar to the one in osc-rdma is used to loop over the
contiguous blocks of each datatype. The operation is
performed using multiple Portals4 calls.
This commit ensures that individual operations do not exceed the
max atomic size or the max message size supported by the device.
Signed-off-by: Todd Kordenbrock <thkgcode@gmail.com>
add padding so the memory allocated by MPI_Win_allocate_shared()
is 64 bytes aligned.
Thanks Joseph Schuchart for the bug report
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
This commit fixes a number of threading issues discovered in
osc/pt2pt. This includes:
- Lock the synchronization object not the module in osc_pt2pt_start.
This fixes a race between the start function and processing post
messages.
- Always lock before calling cond_broadcast. Fixes a race between
the waiting thread and signaling thread.
- Make all atomically updated values volatile.
- Make the module lock recursive to protect against some deadlock
conditions. Will roll this back once the locks have been
re-designed.
- Mark incoming complete *after* completing an accumulate not
before. This was causing an incorrect answer under certain
conditions.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Using MPI_MINLOC or MPI_MAXLOC with the following data types
leads to data corruption:
* MPI_DOUBLE_INT
* MPI_LONG_INT
* MPI_SHORT_INT
* MPI_LONG_DOUBLE_INT
Detect this print a error message and abort.
This workaround should be removed once the following issue is resolved:
* https://github.com/open-mpi/ompi/issues/1666
Signed-off-by: Joshua Hursey <jhursey@us.ibm.com>
* When using `MPI_Put` with `MPI_Win_lock_all` a hang is possible since
the `put` is waiting on `eager_send_active` to become `true` but
that variable might not be reset in the case of `MPI_Win_lock_all`
depending on other incoming events (e.g., `post` or ACKs of lock
requests.
Signed-off-by: Joshua Hursey <jhursey@us.ibm.com>
* When using `MPI_Lock`/`MPI_Unlock` with `MPI_Get` and non-contiguous
datatypes is is possible that the unlock finishes too early before
the data is actually present in the recv buffer.
* We need to wait for the irecv to complete before unlocking the target.
This commit waits for the outgoing fragment counts to become equal
before unlocking.
Signed-off-by: Joshua Hursey <jhursey@us.ibm.com>
* If the user uses PSCW synchronization after a Fence then the previous
epoch is not reset which can cause the PSCW to transfer data before
it is ready leading to wrong answers.
* This commit resets the `eager_send_active` in the start call.
Signed-off-by: Joshua Hursey <jhursey@us.ibm.com>
Instead of ompi_datatype_get_extent(), use ompi_datatype_get_true_extent()
to get the local and remote lower bound. For derived types like
subarray, true_lb is the correct offset for RDMA operations.
Instead of ompi_datatype_get_extent(), use ompi_datatype_get_true_extent()
to get the origin and target lower bound. For derived types like
subarray, true_lb is the correct offset for RDMA operations. Also,
instead of the extent use the size of the datatype.
This commit fixes a typo in compare-and-swap when retrieving the
memory region associated with a displacement. It was erroneously 8
bytes instead of the datatype size. This can cause an incorrect RMA
range error when the compare-and-swap is less than 4 bytes from the
end of the region.
Fixedopen-mpi/ompi#2080
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit adds support for using network AMOs for MPI_Accumulate,
MPI_Fetch_and_op, and MPI_Compare_and_swap. This support is only
enabled if the ompi_single_intrinsic info key is specified or the
acc_single_interinsic MCA variable is set. This configuration
indicates to this implementation that no long accumulates will be
performed since these do not currently mix with the AMO
implementation.
This commit also cleans up the code somwhat. This includes removing
unnecessary struct keywords where the type is also typedef'd.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit cleans up some code in the passive target path. The code
used the buffered frag control send path but it is more appropriate to
use the unbuffered one. This avoids checking structures that are
should not be in use in this path.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes an ordering bug in the code that keeps track of all
attached memory windows. The code is intended to keep the memory
regions sorted but was often inserting at the wrong index. Thanks to
Christoph Niethammer for reporting the issue. The reproducer will be
added to nightly MTT testing.
Fixesopen-mpi/ompi#2012
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
It is possible for another thread to process a lock ack before the
peer is set as locked. In this case either setting the locked or the
eager active flag might clobber the other thread. To address this the
flags have been made volatile and are set atomically. Since there is
no a opal_atomic_or or opal_atomic_and function just use cmpset for
now.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes some bugs uncovered during thread testing of
2.0.1rc1. With these fixes the component is running cleanly with
threads.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit changes the sematics of ompi request callbacks. If a
request's callback has freed or re-posted (using start) a request
the callback must return 1 instead of OMPI_SUCCESS. This indicates
to ompi_request_complete that the request should not be modified
further. This fixes a race condition in osc/pt2pt that could lead
to the req_state being inconsistent if a request is freed between
the callback and setting the request as complete.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
The original lock_all algorithm in osc/pt2pt sent a lock message to
each peer in the communicator even if the peer is never the target of
an operation. Since this scales very poorly the implementation has
been replaced by one that locks the remote peer on first communication
after a call to MPI_Win_lock_all.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes an issue that can occur if a target gets overwhelmed with
requests. This can cause osc/pt2pt to go into deep recursion with a stack
like req_complete_cb -> ompi_osc_pt2pt_callback -> start -> req_complete_cb
-> ... . At small scale this is fine as the recursion depth stays small but
at larger scale we can quickly exhaust the stack processing frag requests.
To fix the issue the request callback now simply puts the request on a
list and returns. The osc/pt2pt progress function then handles the
processing and reposting of the request.
As part of this change osc/pt2pt can now post multiple fragment receive
requests per window. This should help prevent a target from being overwhelmed.
Signed-off-by: Nathan Hjelm <hjelmn@me.com>
It is possible for the start call to complete the requests. For this
reason the module rdma_frag field should be filled in before start is
called. If the request completes the completion callback will reset
the rdma_frag field to NULL. Fixes a bug discovered by @tkordenbrock.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit expands the OPAL_THREAD macros to include 32- and 64-bit
atomic swap. Additionally, macro declararations have been updated to
include both OPAL_THREAD_* and OPAL_ATOMIC_*. Before this commit the
former was used with add and the later with cmpset.
Signed-off-by: Nathan Hjelm <hjelmn@me.com>
This commit fixes a bug in the RDMA compare-and-swap implementation
that caused the origin value to always be written even if the compare
should have failed.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Need to increment the total size after checking the local offset not
before. This typo causes large allocations with MPI_Win_allocate() to
fail.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
* mpi/start: fix bugs in cm and ob1 start functions
There were several problems with the implementation of start in Open
MPI:
- There are no checks whatsoever on the state of the request(s)
provided to MPI_Start/MPI_Start_all. It is erroneous to provide an
active request to either of these calls. Since we are already
looping over the provided requests there is little overhead in
verifying that the request can be started.
- Both ob1 and cm were always throwing away the request on the
initial call to start and start_all with a particular
request. Subsequent calls would see that the request was
pml_complete and reuse it. This introduced a leak as the initial
request was never freed. Since the only pml request that can
be mpi complete but not pml complete is a buffered send the
code to reallocate the request has been moved. To detect that
a request is indeed mpi complete but not pml complete isend_init
in both cm and ob1 now marks the new request as pml complete.
- If a new request was needed the callbacks on the original request
were not copied over to the new request. This can cause osc/pt2pt
to hang as the incoming message callback is never called.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
* osc/pt2pt: add request for gc after starting a new request
Starting a new receive may cause a recursive call into the pt2pt
frag receive function. If this happens and the prior request is
on the garbage collection list it could cause problems. This commit
moves the gc insert until after the new request has been posted.
Signed-off-by: Nathan Hjelm <hjelmn@me.com>
* Remodel the request.
Added the wait sync primitive and integrate it into the PML and MTL
infrastructure. The multi-threaded requests are now significantly
less heavy and less noisy (only the threads associated with completed
requests are signaled).
* Fix the condition to release the request.
This commit fixes a bad synchronization detection bug that occurs when
mixing MPI_Win_fence() and MPI_Win_lock(). If no communication has
occurred in the fence epoch it is safe to just clear the all_sync
object (it was set up by fence).
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes a bug that occurs when ranks are either not mapped
evenly or by something other than core.
Fixesopen-mpi/ompi#1599
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Before this commit, a same PML tag may be used for distinct
communications for long messages. For example, consider a condition
where rank A calls ```MPI_PUT``` targeting rank B and rank B calls
```MPI_GET``` targeting rank A simultaneously.
A PML tag for the ```MPI_PUT``` is acquired on rank A and is used
for the long-message communication from rank A to rank B.
A PML tag for the ```MPI_GET``` is acquired on rank B and is used
for the long-message communication from rank A to rank B.
These two tags may become a same value because they are managed
independently on each rank. This will cause a data corruption.
This commit separates the tag used in a single RMA communication
call, one for communication from an origin to a target, and one
for communication from a target to an origin. A "base" tag
is acquired using ```get_tag``` function and PML tag is caluculated
from the base tag by ```tag_to_target``` and ```tag_to_origin```
function.
Fix CID 1324726 (#1 of 1): Free of address-of expression (BAD_FREE):
Indeed, if a lock conflicts with the lock_all we will end up trying to
free an invalid pointer.
Fix CID 1328826 (#1 of 1): Dereference after null check (FORWARD_NULL):
This was intentional but it would be a good idea to check for
module->comm being non_NULL to be safe. Also cleaned out some checks
for NULL before free().
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes several bugs identified by @ggouaillardet and MTT:
- Fix SEGV in long send completion caused by missing update to the
request callback data.
- Add an MPI_Barrier to the fence short-cut. This fixes potential
semantic issues where messages may be received before fence is
reached.
- Ensure fragments are flushed when using request-based RMA. This
allows MPI_Test/MPI_Wait/etc to work as expected.
- Restore the tag space back to 16-bits. It was intended that the
space be expanded to 32-bits but the required change to the
fragment headers was not committed. The tag space may be expanded
in a later commit.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes several bugs identified by a new multi-threaded RMA
benchmarking suite. The following bugs have been identified and fixed:
- The code that signaled the actual start of an access epoch changed
the eager_send_active flag on a synchronization object without
holding the object's lock. This could cause another thread waiting
on eager sends to block indefinitely because the entirety of
ompi_osc_pt2pt_sync_expected could exectute between the check of
eager_send_active and the conditon wait of
ompi_osc_pt2pt_sync_wait.
- The bookkeeping of fragments could get screwed up when performing
long put/accumulate operations from different threads. This was
caused by the fragment flush code at the end of both put and
accumulate. This code was put in place to avoid sending a large
number of unexpected messages to a peer. To fix the bookkeeping
issue we now 1) wait for eager sends to be active before stating
any large isend's, and 2) keep track of the number of large isends
associated with a fragment. If the number of large isends reaches
32 the active fragment is flushed.
- Use atomics to update the large receive/send tag counters. This
prevents duplicate tags from being used. The tag space has also
been updated to use the entire 16-bits of the tag space.
These changes should also fixopen-mpi/ompi#1299.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit adds code to handle large unaligned gets. There are two
possible code paths for these transactions:
1) The remote region and local region have the same alignment. In
this case the get will be broken down into at most three get
transactions: 1 transaction to get the unaligned start of the region
(buffered), 1 transaction to get the aligned portion of the region,
and 1 transaction to get the end of the region.
2) The remote and local regions do not have the same alignment. This
should be an uncommon case and is not optimized. In this case a
buffer is allocated and registered locally to hold the aligned data
from the remote region. There may be cases where this fails (low
memory, can't register memory). Those conditions are unlikely and
will be handled later.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
If atomics are not globally visible (cpu and nic atomics do not mix)
then a btl endpoint must be used to access local ranks. To avoid
issues that are caused by having the same region registered with
multiple handles osc/rdma was updated to always use the handle for
rank 0. There was a bug in the update that caused osc/rdma to continue
using the local endpoint for accessing the state even though the
pointer/handle are not valid for that endpoint. This commit fixes the
bug.
Fixesopen-mpi/ompi#1241.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Optimizing put aggregation in the presence of threads will require a
redesign of the code. For now just ensure that put aggregation is
turned off when MPI_THREAD_MULTIPLE is enabled.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
A bus error occurs in sm OSC under the following conditions.
- sparc64 or any other architectures which need strict alignment.
- `MPI_WIN_POST` or `MPI_WIN_START` is called for a window created
by sm OSC.
- The communicator size is odd and greater than 3.
The lines 283-285 in current `ompi/mca/osc/sm/osc_sm_component.c` has
the following code.
```c
module->global_state = (ompi_osc_sm_global_state_t *) (module->segment_base);
module->node_states = (ompi_osc_sm_node_state_t *) (module->global_state + 1);
module->posts[0] = (uint64_t *) (module->node_states + comm_size);
```
The size of `ompi_osc_sm_node_state_t` is multiples of 4 but not
multiples of 8. So if `comm_size` is odd, `module->posts[0]` does
not aligned to 8. This causes a bus error when accessing
`module->posts[i][j]`.
This patch fixes the alignment of `module->posts[0]` by setting
`module->posts[0]` first.
NOTE: Building with external pmix *requires* that you also build with external libevent and hwloc libraries. Detect this at configure and error out with large message if this requirement is violated.
Closes#1204 (replaces it)
Fixes#1064
A previous commit updated the one-sided code to register the state
region only once. This created an issue when using the scratch lock
with fetching atomics. In this case on any rank that isn't local rank
0 the module->state_handle is NULL. This commit fixes the issue by
removing the scratch lock and using a fragment pointer instead.
Fixesopen-mpi/ompi#1290
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
`MPI_WIN_TEST` must update the `flag` parameter to 0 when not all
origin processes called `MPI_WIN_COMPLETE`. But sm OSC doesn't.
If the caller initialize the `flag` argument to a non-0 value,
the caller will receive the non-0 `flag` value.
This commit changes the way ompi_proc_t's are retained/released by
ompi_group_t's. Before this change ompi_proc_t's were retained once
for the group and then once for each retain of a group. This method
adds unnecessary overhead (need to traverse the group list each time
the group is retained) and causes problems when using an async
add_procs.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
There were two bugs in osc/rdma when using threads:
- Deadlock is ompi_osc_rdma_start_atomic. This occurs because
ompi_osc_rdma_frag_alloc is called with the module lock. To fix the
issue the module lock is now recursive. In the future I will add a
new lock to protect just the current rdma fragment.
- Do not drop the lock in ompi_osc_rdma_frag_alloc when calling
ompi_osc_rdma_frag_complete. Not only is it not needed but dropping
the lock at this point can cause a competing thread to mess up the
state.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes a bug that can occur on Cray Gemini networks. If
multiple registrations are used for the local state then we looks the
atomicity guarantees. To avoid issues like this use only a single
registration handle for all local state on a node.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes the following:
- CIDs 1328491, 1328492: Dead code caused by typos in a prior
commit.
- Fix the calculation of dynamic memory regions. This was causes
incorrect RMA range errors when accessing the last partial page of
an attachment.
- Fix a SEGV when using dynamic memory windows with local state (all
processes on the same node).
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes a bug that occurs when a post message comes in when
sending complete messages or while waiting for all outgoing messages
to flush. In that case the post message might get incorrecly
associated with the ending sync object.
References open-mpi/ompi#1012
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes several bugs in the osc/rdma component:
- Complete aggregated requests immediately. Completion of RMA
requests indicates local completion anyway. This fixes a hang in
the c_reqops test.
- Correctly mark Rget_accumulate requests.
- Set the local base flag correctly on the local peer.
- Clear or set the no locks flag on the window if the value is
changed by MPI_Win_set_info.
- Actually update the target when using MPI_OP_REPLACE.
Fixesopen-mpi/ompi#1010
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes an issue identified by @rolfv. The local peer was
not being correctly initialized when running with a single process on
a node.
This fixesopen-mpi/ompi#1010
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Fix code paths that didn't convert the MPI datatype to the
corresponding Portals4 datatype.
Thanks to Nicolas Chevalier (@shawone) for finding this bug and
submitting a patch.
Fixed CID 1269712, 1269709, 1269706, 1269703, 1269694: Logically dead code
Remove extra NULL check as OMPI_OSC_PT2PT_REQUEST_ALLOC can never set the
request to NULL.
Fixes CID 1269668: Unchecked return value
False positive. Add (void) to indicate we do not care about the return code
from opal_hash_table_get_uint32.
Fixes CID 1324726: Free of address-of expression
Do not free lock if it was not allocated.
Fixes CID 1269658: Free of address-of expression
Never will happen but because op is always a built-in op there is no
reason to retain/release it anyway.
Signed-off-by: Nathan Hjelm <hjelmn@me.com>
In the default mode of operation, the Portals4 components support
dynamic add_procs().
The Portals4 components have two alternate modes (flow control and
logical-to-physical) that require knowledge of all procs at startup.
In these modes, mtl-portals4 sets the MCA_MTL_BASE_FLAG_REQUIRE_WORLD
flag and btl-portals4 sets the MCA_BTL_FLAGS_SINGLE_ADD_PROCS flag
to tell the PML that we need all the procs in one add_procs() call.
The osc/sm component was using a simple counter to determine if all
expected posts had arrived to start a PSCW access epoch. This is
incorrect as a post may arrive from a peer that isn't part of the
current start group. There are many ways this could have been fixed.
This commit adds an n^2 bitmap. When a process posts it sets a bit in
the bitmap associated with the access rank to indicate the post is
complete. The access rank checks for and clears the bits associated
with all the processes in the start group.
The bitmap requires comm_size ^ 2 bits of space. This should be
managable as most nodes have relatively small numbers of processes. If
this changes another algorigthm can be implemented.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Fix CID 1324733: Null pointer dereferences (FORWARD_NULL)
Fix CID 1324734: Null pointer dereferences (FORWARD_NULL)
Fix CID 1324735: Null pointer dereferences (FORWARD_NULL)
Fix CID 1324736: Null pointer dereferences (FORWARD_NULL)
Fix CID 1324737: Null pointer dereferences (FORWARD_NULL)
Fix CID 1324751: Memory - illegal accesses (USE_AFTER_FREE)
Fix CID 1324750: (USE_AFTER_FREE)
Fix CID 1324749: Memory - corruptions (USE_AFTER_FREE)
Fix CID 1324748: Memory - illegal accesses (USE_AFTER_FREE)
Fix CID 1324747: (USE_AFTER_FREE)
Fix CID 1324746: Memory - corruptions (USE_AFTER_FREE)
Add missing return on an error path.
Fix CID 1324745: Code maintainability issues (UNUSED_VALUE)
Ignore return code from barrier. It was not being used anyway.
Fix CID 1324738: Null pointer dereferences (FORWARD_NULL)
Fix CID 1324741: Null pointer dereferences (REVERSE_INULL)
module->selected_btl can not be NULL in osc/rdma during normal
operation. Removed the unnecessary NULL check.
Fix CID 1324752: Memory - illegal accesses (USE_AFTER_FREE)
Move ompi_osc_pt2pt_module_lock_remove to before the lock is freed.
Fix CID 1324744: Uninitialized variables (UNINIT)
Fix CID 1324743: Uninitialized variables (UNINIT)
This array is not used unitialized but there is no reason not to use
calloc here to silence the warning.
The following CID is a false positive: 1324742. I will mark it such in
coverity.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit adds support for performing one-sided operations over
supported hardware (currently Infiniband and Cray Gemini/Aries). This
component is still undergoing active development.
Current features:
- Use network atomic operations (fadd, cswap) for implementing
locking and PSCW synchronization.
- Aggregate small contiguous puts.
- Reduced memory footprint by storing window data (pointer, keys,
etc) at the lowest rank on each node. The data is fetched as each
process needs to communicate with a new peer. This is a trade-off
between the performance of the first operation on a peer and the
memory utilization of a window.
TODO:
- Add support for the accumulate_ops info key. If it is known that
the same op or same op/no op is used it may be possible to use
hardware atomics for fetch-and-op and compare-and-swap.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit updates osc/pt2pt to allocate peer object as they are
needed rather than all at once. Additionally, to help improve the
memory footprint a new synchronization structure has been added.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit modifies the ompi_group_t union/difference code to compare/copy the
raw group values. This will either be a ompi_proc_t or a sentinel value. This
commit also adds helper functions to convert between opal process names and
sentinel values.
Signed-off-by: Nathan Hjelm <hjelmn@me.com>
- MPI_Compare_and_swap
- MPI_Fetch_and_op
- MPI_Raccumulate
- MPI_Win_detach
Thanks to Michael Knobloch and Takahiro Kawashima for bringing this
to our attention
Portals4 supports atomic ops on datatypes less than or equal to
max_fetch_atomic_size bytes. This commit fixes a bug that required
the datatype to be less than max_fetch_atomic_size bytes.
