1
1
Граф коммитов

8 Коммитов

Автор SHA1 Сообщение Дата
Harumi Kuno
1bc3dab118 Add comments about order of close ops
Per suggestion of @awlauria, added some comments about
the need to free ep before resources it points to.

Signed-off-by: Harumi Kuno <harumi.kuno@hpe.com>
2020-03-07 14:08:39 -07:00
harumi kuno
3095fabf94 Fix mca_btl_ofi_finalize clean-up logic
This fix is from John L. Byrne (john.l.byrne@hpe.com).

When OFI Libfabric binds objects to endpoints, before the object can
be successfully closed, the endpoint must first be freed.  For scalable
endpoints, objects can also be bound to transmit and receive contexts,
and for objects that are bound to contexts, we need to first free the
contexts before freeing the endpoint. We also need to clear the memory
registration cache.

If we don't clean up properly, then fi\_close may not be able to close
the domain because the dom will have a non-zero ref count.

Signed-off-by: harumi kuno <harumi.kuno@hpe.com>
2020-03-04 17:51:08 -07:00
Aravind Gopalakrishnan
786e686d43 btl/ofi: Fix valgrind complaints on uninitialized pointer use
It doesn't seem like the BTL was using uninitialized pointer. But simply
setting the rcache pointer to NULL after destroying it makes the valgrind
errors go away.

Fixes Issue #6345

Signed-off-by: Aravind Gopalakrishnan <Aravind.Gopalakrishnan@intel.com>
2019-02-01 14:03:23 -08:00
Brian Barrett
e9e4d2a4bc Handle asprintf errors with opal_asprintf wrapper
The Open MPI code base assumed that asprintf always behaved like
the FreeBSD variant, where ptr is set to NULL on error.  However,
the C standard (and Linux) only guarantee that the return code will
be -1 on error and leave ptr undefined.  Rather than fix all the
usage in the code, we use opal_asprintf() wrapper instead, which
guarantees the BSD-like behavior of ptr always being set to NULL.
In addition to being correct, this will fix many, many warnings
in the Open MPI code base.

Signed-off-by: Brian Barrett <bbarrett@amazon.com>
2018-10-08 16:43:53 -07:00
Jeff Squyres
8f2620d3af misc: compiler warning fixes
A variety of small compiler warning fixes.  The 2 PMIx fixes are
already committed upstream.

Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
2018-09-15 06:04:13 -07:00
Thananon Patinyasakdikul
080115d440 btl/ofi: Added 2 side communication support.
The 2 sided communication support is added for non-tagmatching provider
to take advantage of this BTL and PML OB1. The current state is
"functional" and not optimized for performance.

Two sided support is disabled by default and can be turned on by mca
parameter: "mca_btl_ofi_mode".

Signed-off-by: Thananon Patinyasakdikul <thananon.patinyasakdikul@intel.com>
2018-08-03 12:30:03 -07:00
Thananon Patinyasakdikul
dae3c9447c btl/ofi: add scalable endpoint support.
This commit add support for scalable endpoint to enhance multithreaded
application performance. The BTL will detect the support from ofi
provider and will fallback to normal usage of scalable endpoint is not
supported.

NEW MCA parameters:
- mca_btl_ofi_disable_sep: force the btl to not use scalable endpoint.
- mca_btl_ofi_num_contexts_per_module: number of communication context
  to create (should be the same as number of thread).

Signed-off-by: Thananon Patinyasakdikul <thananon.patinyasakdikul@intel.com>
2018-06-14 15:44:29 -07:00
Thananon Patinyasakdikul
6efb8069ec new btl/ofi: RDMA only btl using libfabric.
This commit added new transport layer to be used with osc rdma module.
This BTL provides put, get, atomic and fetch atomic operations. It can
be used with multiple hardware vendors as long as they have their
provider under Libfabric and have the right capabilities.

Signed-off-by: Thananon Patinyasakdikul <thananon.patinyasakdikul@intel.com>
2018-06-01 15:22:04 -07:00