This commit fixes an issue seem with some older versions of gcc
(verified to occur in gcc 6.x) where on x86_64 systems the
acquire memory barrier in C11 atomics acts as a no-op. On these
systems the three memory barriers should all be equivalent.
This is related to the error fixed in open-mpi/ompi@30119ee.
References #6655.
Signed-off-by: Nathan Hjelm <hjelmn@google.com>
Remove compatibility code for multiple versions of BTL_IN_OPAL,
BTL_VERSION, and RCACHE_VERSION. This stuff was really only necessary
when we were actively swapping code between multiple release branches
that had large variations in core OMPI infrastructure. These large
variations have now been around for quite a while, so the need for
this "compat" layer is significantly reduced. It hasn't been removed
simply because a few of the "compat" names a slightly more friendly
than the real names (e.g., the SEND/RECV/PUT names).
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
This commit fixes an error in the 32-bit compare-and-swap atomic support
for Aries networks. The code was incorrectly using the non-fetching
version of cswap which was causing the routing to return
OPAL_ERR_BAD_ARG.
Signed-off-by: Nathan Hjelm <hjelmn@cs.unm.edu>
The new routine transfers the data asynchronously from the source PE to all
PEs in the OpenSHMEM job. The routine returns immediately. The source and
target buffers are reusable only after the completion of the routine.
After the data is transferred to the target buffers, the counter object
is updated atomically. The counter object can be read either using atomic
operations such as shmem_atomic_fetch or can use point-to-point synchronization
routines such as shmem_wait_until and shmem_test.
Signed-off-by: Mikhail Brinskii <mikhailb@mellanox.com>
This link-back seems to be breaking OMPI for some reason. I'm not sure we need it in PMIx anyway, but we'll investigate over there.
Signed-off-by: Ralph Castain <rhc@pmix.org>
The first category of issue I'm addressing is that recent code changes
seem to only consider -cpu-set as a binding option. Eg a command like
this
% mpirun -np 2 --report-bindings --use-hwthread-cpus \
--bind-to cpulist:ordered --map-by hwthread --cpu-set 6,7 hostname
which just round robins over the --cpu-set list.
Example output which seems fine to me:
> MCW rank 0: [..../..B./..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
> MCW rank 1: [..../...B/..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
It should also be possible though to pass a --cpu-set to most other
map/bind options and have it be a constraint on that binding. Eg
% mpirun -np 2 --report-bindings \
--bind-to hwthread --map-by hwthread --cpu-set 6,7 hostname
% mpirun -np 2 --report-bindings \
--bind-to hwthread --map-by ppr:2:node,pe=2 --cpu-set 6,7,12,13 hostname
The first command above errors that
> Conflicting directives for mapping policy are causing the policy
> to be redefined:
> New policy: RANK_FILE
> Prior policy: BYHWTHREAD
The error check in orte_rmaps_rank_file_open() is likely too aggressive.
The intent seems to be that any option like "--map-by whatever" will
check to see if a rankfile is in use, and report that mapping via rmaps
and using an explicit rankfile is a conflict.
But the check has been expanded to not just check
NULL != orte_rankfile
but also errors out if
(NULL != opal_hwloc_base_cpu_list &&
!OPAL_BIND_ORDERED_REQUESTED(opal_hwloc_binding_policy))
which seems to be only recognizing -cpu-set as a binding option and
ignoring -cpu-set as a constraint on other binding policies.
For now I've changed the
NULL != opal_hwloc_base_cpu_list
to
OPAL_BIND_TO_CPUSET == OPAL_GET_BINDING_POLICY(opal_hwloc_binding_policy)
so it hopefully only errors out if -cpu-set is being used as a binding
policy. Whether I did that right or not it's enough to get to the next
stage of testing the example commands I have above.
Another place similar logic is used is hwloc_base_frame.c where it has
/* did the user provide a slot list? */
if (NULL != opal_hwloc_base_cpu_list) {
OPAL_SET_BINDING_POLICY(opal_hwloc_binding_policy, OPAL_BIND_TO_CPUSET);
}
where it used to (long ago) only do that if
!OPAL_BINDING_POLICY_IS_SET(opal_hwloc_binding_policy)
I think the new code is making it impossible to use --cpu-set as anything
other than a binding policy.
That brings us past the error detection and into the real functionality, some of
which has been stripped out, probably in moving to hwloc-2:
% mpirun -np 2 --report-bindings \
--bind-to hwthread --map-by hwthread --cpu-set 6,7 hostname
> MCW rank 0: [B.../..../..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
> MCW rank 1: [.B../..../..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
The rank_by() function in rmaps_base_ranking.c makes an array out of objects
returned from
opal_hwloc_base_get_obj_by_type(,,,i,)
which uses df_search(). That function changed quite a bit from hwloc-1 to 2
but it used to include a check for
available = opal_hwloc_base_get_available_cpus(topo, start)
which is where the bitmask from --cpu-set goes. And it used to skip objs that
had hwloc_bitmap_iszero(available).
So I restored that behavior in ds_search() by adding a "constrained_cpuset" to
replace start->cpuset that it was otherwise processing. With that change in
place the first command works:
% mpirun -np 2 --report-bindings \
--bind-to hwthread --map-by hwthread --cpu-set 6,7 hostname
> MCW rank 0: [..../..B./..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
> MCW rank 1: [..../...B/..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
The other command uses a different path though that still ignored the
available mask:
% mpirun -np 2 --report-bindings \
--bind-to hwthread --map-by ppr:2:node:pe=2 --cpu-set 6,7,12,13 hostname
> MCW rank 0: [BB../..../..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
> MCW rank 1: [..BB/..../..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
In bind_generic() the code used to call
opal_hwloc_base_find_min_bound_target_under_obj() which used
opal_hwloc_base_get_ncpus(), and that's where it would
intersect objects with the available cpuset and skip over ones
that were't available. To match the old behavior I added a few
lines in bind_generic() to skip over objects that don't intersect
the available mask. After that we get
% mpirun -np 2 --report-bindings \
--bind-to hwthread --map-by ppr:2:node:pe=2 --cpu-set 6,7,12,13 hostname
> MCW rank 0: [..../..BB/..../..../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
> MCW rank 1: [..../..../..../BB../..../..../..../..../..../..../..../....][..../..../..../..../..../..../..../..../..../..../..../....]
I think the above changes are improvements, but I don't feel like they're
comprehensive. I only traced through enough code to fix the two specific
bugs I was dealing with.
Signed-off-by: Mark Allen <markalle@us.ibm.com>
... to avoid using an architecture name macro in
`opal/mca/timer/linux/timer_linux_component.c`.
The function name `opal_sys_timer_freq` is also changed for
consistency with `opal_sys_timer_get_cycles`.
Signed-off-by: KAWASHIMA Takahiro <t-kawashima@jp.fujitsu.com>
... in the case of `OPAL_GCC_INLINE_ASSEMBLY == 0`
In this case, `OPAL_HAVE_SYS_TIMER_GET_CYCLES` should be 0 because
the `opal_sys_timer_get_cycles` function is not defined.
The history:
1. Before 8d4175ad89, `OPAL_HAVE_SYS_TIMER_GET_CYCLES` was 0.
2. In 8d4175ad89, adf92d6237, adf92d6237, and c62ce1593a,
`OPAL_HAVE_SYS_TIMER_GET_CYCLES` was changed to 1 by introducing
`opal/asm/base/*.asm`.
3. In ebce88b7ad, `opal/asm/base/*.asm` were removed.
Signed-off-by: KAWASHIMA Takahiro <t-kawashima@jp.fujitsu.com>
This is mostly based off recent UCX additions to their patcher:
https://github.com/openucx/ucx/pull/2703
They added triggers for
* mmap when (flags & MAP_FIXED) && (addr != NULL)
* shmat when (shmflg & SHM_REMAP) && (shmaddr != NULL)
Beyond that I noticed they already had a trigger for
* madvise when (advice == MADV_FREE)
that we didn't so I added that.
And the other main thing is we didn't really have shmat/shmdt
active for some systems because we only had a path for
syscall(SYS_shmdt, ) but we needed to also have a path for
syscall(SYS_ipc, IPCOP_shmdt, ) and same for shmat.
Signed-off-by: Mark Allen <markalle@us.ibm.com>
Avoid printing an error message about ENOTCONN return codes from
getpeername() when handling an incoming connection request. At
this point in the receive state machine, the remote process has
been verified to be a valid OMPI instance. In all-to-all startup
at 4k rank scale, we're seeing this error message when the remote
side drops the connection because it realizes it's the "loser"
in the connection race. We were already doing all the right things,
other than printing a scary error message. So skip the error
message and call it good.
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
free the component mpool in mca_btl_vader_component_close()
and after freeing soem objects that depend on it such as
mca_btl_vader_component.vader_frags_user
Thanks Christoph Niethammer for reporting this.
Refs. open-mpi/ompi#6524
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
- there was a set of UCX related issues reported which caused
by mmap API hooks conflicts. We added diagnostic of such
problems to simplify bug-resolving pipeline
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
I think the strncat() calls here need to be of the form
strncat(str, new_str_to_add, len - strlen(new_str_to_addstr) - 1);
since in the OMPI calls len is being used as total number of bytes
in str.
strncat(dest,src,n) on the other hand is documented as writing up to
n chars from the incoming string plus 1 for the null, for n+1 total
bytes it can write.
Signed-off-by: Mark Allen <markalle@us.ibm.com>
Use $(AM_CPPFLAGS) in $(usnic_btl_run_tests_CPPFLAGS) so that we don't
have to replicate hard-coded values.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
OPAL_XLC_INLINE_ASSEMBLY was removed in commit ebce88b7ad.
Removing dead code, which also fixes a compiler warning.
Signed-off-by: Nysal Jan K.A <jnysal@in.ibm.com>
do define the OMPI_LIBMPI_NAME macro via the CPPFLAGS.
The issue occurs when Open MPI is configured with
--enable-opal-btl-usnic-unit-tests
Thanks George Marselis for reporting this issue
Refs. open-mpi/ompi#6441
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
When Slurm is built against PMIx, some installations place a copy of the
PMIx library that Slurm is linking against in the Slurm PMI location.
Current configury ignores that location. The desired behavior is to look
for a PMIx lib in that location when --with-pmi is given. If the user
also specifies --with-pmix and gives a different location, then override
anything previously found and look for it where the user directed.
Signed-off-by: Ralph Castain <rhc@pmix.org>
Place the content of common_ucx_int.h back to the common_ucx.h and
include common_ucx_wpool.h explicitly.
Signed-off-by: Artem Polyakov <artpol84@gmail.com>
Worker Pool is an object containing/managing a set of UCX workers
and providing access to those workers through a smal interface
to allow Multi-Threaded applicatoins to access multiple HW contexts.
Signed-off-by: Artem Polyakov <artpol84@gmail.com>
We've been fighting the battle of trying to create a regex generator and
parser that can handle arbitrary hostname schemes - without long-term
success. The worst of it is that there is no way of checking to see if
the computed regex is correct short of parsing it and doing a
character-by-character comparison with the original string. Ugh...there
has to be a better solution.
One option is to investigate using 3rd-party regex libraries as
those are coming from communities whose sole focus is resolving that
problem. However, someone would need to spend the time to investigate
it, and we'd have to find a license-friendly implementation.
Another option is to quit beating our heads against the wall and just
compress the information. It won't be as much of a reduction, but we
also won't keep hitting scenarios where things break. In this case, it
seems that "perfection" is definitely the enemy of "good enough".
This PR implements the compression option while retaining the
possibility of people adding regex-generating components. The
compression code used in ORTE is consolidated into the opal/compress
framework. That framework currently held bzip and gzip components for
use in compressing checkpoint files - since we no longer support C/R, I
have .opal_ignore'd those components.
However, I have left the original framework APIs alone in case someone
ever decides to redo C/R. The APIs of interest here are added to the
framework - specifically, the "compress_block" and "decompress_block"
functions. I then moved the ORTE zlib compression code into a new
component in this framework.
Unfortunately, the framework currently is a single-select one - i.e.,
only one active component at a time. Since I .opal_ignore'd the other
two and made the priority of zlib high, this isn't a problem. However,
if someone wants to re-enable bzip/gzip or add another component, they
might need to transition opal/compress to a multi-select framework.
Included changes:
* Consolidate the compression code into the opal/compress framework
* Move the ORTE zlib compression code into a new opal/compress/zlib
component
* Ignore the bzip and gzip components in opal/compress framework
* Add a "compress_base_limit" MCA param to set the threshold above which
we compress data - defaults to 4096 bytes
* Delete stale brucks and rcd components from orte/grpcomm framework
* Delete the orte/regx framework
* Update the launch system to use opal/compress instead of string regex
* Provide a default module if no zlib is available
* Fix some misc multi-node issues
* Properly generate the nidmap in response to a "connection warmup"
message so the remote daemon knows the children it needs to launch.
* Remove stale references to orte_node_regex
* opal_byte_object_t's are not OPAL objects - properly release allocated
memory.
* Set the topology
* Currently only handling homogeneous case
* Update the compress framework files to conform
* Consolidate open/close into one "frame" file. Ensure we open/close the
framework
Signed-off-by: Ralph Castain <rhc@pmix.org>
It never lived up to its purpose (and has caused amorphous indirect
errors such as https://github.com/open-mpi/ompi/issues/2519), so
delete it.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Update the OPAL_CHECK_OFI configury macro:
- Make it safe to call the macro multiple times:
- The checks only execute the first time it is invoked
- Subsequent invocations, it just emits a friendly "checking..."
message so that configure output is sensible/logical
- With the goal of ultimately removing opal/mca/common/ofi, rename the
output variables from OPAL_CHECK_OFI to be
opal_ofi_{happy|CPPFLAGS|LDFLAGS|LIBS}.
- Update btl/ofi, btl/usnic, and mtl/ofi for these new conventions.
- Also, don't use AC_REQUIRE to invoke OPAL_CHECK_OFI because that
causes the macro to be invoked at a fairly random time, which makes
configure stdout confusing / hard to grok.
- Remove a little left-over kruft in OPAL_CHECK_OFI, too (which
resulted in an indenting change, making the change to
opal_check_ofi.m4 look larger than it really is).
Thanks Alastair McKinstry for the report and initial fix.
Thanks Rashika Kheria for the reminder.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Now that all components that use libibverbs are gone, remove
OPAL_CHECK_VERBS and the confusingly-named OPAL_CHECK_OPENFABRICS
(which really just checked for verbs things -- not all the possible
OpenFabrics APIs/libraries).
The only code left in Open MPI that calls verbs is hwloc -- and that's
just the APIs that takes an IBV device and returns topological
information about it. Since nothing in the Open MPI code base uses
the "ibv_*" API any more, we have no need for this hwloc functionality
so we'll even remove the --with-verbs configure options.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
The verbs and verbs_usnic components are now no longer necessary / no
longer used anywhere in the code base.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
So long BTL openib! After many years of (mostly) faithful service, it
is time to remove the openib BTL. It has been fully replaced by other
components, such as the UCX PML and OFI MTL.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
PMIx is removing the --enable-embedded-libevent and
--enable-embedded-hwloc flags as they are confusing users. Instead, we
will use the --enable-embedded-mode to handle both of these options.
Update the embedded configury to handle it.
Signed-off-by: Ralph Castain <rhc@pmix.org>
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>
... and add `MPI_COMPLEX4`.
This commit changes values of existing `OMPI_DATATYPE_MPI_*` macros.
This change does not affect ABI compatibility of `libmpi.so` and the
like because these values are only used in OMPI internal code.
On the other hand, `ompi_datatype_t::id` values of existing datatypes
are not changed and 73 is newly assigned to for `MPI_COMPLEX4` to
retain ABI compatibility.
Signed-off-by: KAWASHIMA Takahiro <t-kawashima@jp.fujitsu.com>
... and `ompi_mpi_c_short_float_complex` and `ompi_mpi_cxx_sfltcplex`.
These are Open MPI internal variables intended to be defined as
`MPI_SHORT_FLOAT`, `MPI_C_SHORT_FLOAT_COMPLEX`, and
`MPI_CXX_SHORT_FLOAT_COMPLEX` in the future.
`OMPI_DATATYPE_MPI_C_SHORT_FLOAT_COMPLEX` is also required to
support `MPI_COMPLEX4` in the next commit.
Signed-off-by: KAWASHIMA Takahiro <t-kawashima@jp.fujitsu.com>
The type `short float`, which is proposed in ISO/IEC JTC 1/SC 22 WG 14
(C WG), is not supported by most compilers yet. But some compilers
(including gcc 7 for AArch64 and clang 6) support `_Float16`, which
is defined in ISO/IEC TS 18661-3:2015 (ISO/IEC JTC 1/SC 22/WG 14 N1945)
as an extensions for C. If it is detected in `configure`, it is used
as an alternate type of `short float` in Open MPI internal code.
This commit adds a `configure` option `--enable-alt-short-float=TYPE`.
It can be used to specify a type other than `short float` and `_Float16`
as the alternate type.
Signed-off-by: KAWASHIMA Takahiro <t-kawashima@jp.fujitsu.com>
The type `short float` is proposed for the C language in ISO/IEC JTC
1/SC 22 WG 14 (C WG) for mainly IEEE 754-2008 binary16, a.k.a.
half-precision floating point or FP16.
By this commit, `short float` and `short float _Complex` are detected
in `configure` and used in Open MPI internal code. `MPI_SHORT_FLOAT`
and its complex number version are not added yet.
This commit changes values of existing `OPAL_DATATYPE_*` macros.
This change does not affect ABI compatibility of `libmpi.so` and the
like because these values are only used in OPAL and OMPI internal code.
Signed-off-by: KAWASHIMA Takahiro <t-kawashima@jp.fujitsu.com>
Reset ptypes when cloning a datatype in order to prevent
a double free() in the opal_datatype_t destructor.
This fixes a bug introduced in open-mpi/ompi@7c938f070fFixesopen-mpi/ompi#6346
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
The issue was a little complicated due to the internal stack used in the
convertor. The main issue was that in the case where we run out of iov
space to save the raw description of the data while hanbdling a
repetition (loop), instead of saving the current position and bailing out
directly we reading of the next predefined type element. It worked in
most cases, except the one identified by the HDF5 test. However, the
biggest issue here was the drop in performance for all ensuing calls to
the convertor pack/unpack, as instead of handling contiguous loops as a
whole (and minimizing the number of memory copies) we copied data
description by data description.
Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
This commit fixes a bug introduced in
f62d26ddbc. That commit changed how
vader allocates fragment memory from the shared memory
segment. Unfortunately, the values used for the fragment sizes did not
include space for the fragment header. This can cause an overrun of
data from one fragment to the header of the next fragment.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit reverted pr #6199 as it introduced deadlock in some cases.
Also removed the assert as the condition is obsoleted.
Signed-off-by: Thananon Patinyasakdikul <tpatinya@utk.edu>
opal_config_bottom.h can only be #include'd in opal_config.h,
so there is no need to #include "opal_config.h" inside.
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
Some macros defined by the embedded hwloc ends up in opal_config.h
because hwloc configury m4 files are slurped into Open MPI. These
macros are not required here, and they might conflict with an external
hwloc install, so simply #undef them in hwloc/external/external.h
after including <opal_config.h> but before including the external
<hwloc.h>.
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
correctly handle the case in which iovec is full and the
last accessed element of the datatype is the beginning of a loop
Refs. open-mpi/ompi#6285
Thanks Axel Huebl for reporting this
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
correctly free ptypes if the datatype is not pre-defined.
Thanks Axel Huebl for reporting this.
Refs. open-mpi/ompi#6291
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
This commit updates btl/vader to use an mpool for handling all shared
memory allocations (frags, fboxes).
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit adds a new mpool base module type: basic. This module can
be used with an opal_free_list_t to allocate space from a
pre-allocated block (such as a shared memory region). The new module
only supports allocation and is not meant for more dynamic use cases
at this time.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Update the OPAL glue configure code to correctly link the opal/pmix4
component to the hwloc used by OMPI instead of defaulting to the
system-level hwloc. Required a corresponding update to the PMIx hwloc
configure code so we treat hwloc the same way we handle libevent in
embedded scenarios.
Signed-off-by: Ralph Castain <rhc@pmix.org>
This commit fixes the ordering of the teardown for
opal_finalize_util. The installdirs and if frameworks need to come
down before the MCA system.
Fixes#6259
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
The PMIX_MODEX and PMIX_INFO_ARRAY macros were removed from the PMIx 3.1 standard.
Open MPI does not really need them (they are only used to be reported as not supported),
so smply #ifdef protect them to support an external PMIx v3.1
Refs. open-mpi/ompi#6247
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
This commit fixes a bug where add_procs can incorrectly return an
error when going through the dynamic add_procs path. This doesn't
happen normally, only when pml/ob1 is not in use.
References #6201
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
When querying an info value, copy out exactly as many characters as
the caller asked for -- do not artificially truncate the target just
to ensure that it is \0-terminated.
Specifically: do not use opal_string_copy() to copy info values,
because opal_string_copy() will guarantee to \0-terminate the target,
even if it means truncating the target. E.g., if the caller calls
opal_info_get_nolock() with valuelen=5, opal_string_copy() will return
"1234\0" -- which is wrong. This commit fixes the behavior to return
"12345".
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
This commit contains the following changes:
- Remove the unused opal_test_init/opal_test_finalize
functions. These functions are not used by anything in the code
base or MTT. Tests use opal_init_util/opal_finalize_util instead.
- Get rid of gotos in opal_init_util and opal_init. Replaced them
with a cleaner solution.
- Automatically register cleanup functions in init functions. The
cleanup functions are executed in the reverse order of the
initialization functions. The cleanup functions are run in
opal_finalize_util() before tearing down the class system.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
PR #5241 provided an MCA variable to allow multi-threaded opal_progress.
However, it allowed to update the linked list even when multiple threads was
allowed to call opal_progress. This caused a scenario when a more recent thread
could complete it's progress and fail the assert(sync ==
wait_sync_list).
Allowing to update the linked list only for the case when the number of threads
exceeds the threshold fixes the problem.
Signed-off-by: Aravind Gopalakrishnan <Aravind.Gopalakrishnan@intel.com>
remove whitespace around '=' when setting btl_uct_LIBS
Thanks Ake Sandgren for reporting this
Refs. open-mpi/ompi#6173
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
Always use size_t (instead of converting to an uint32_t) in order to
correctly support large datatypes.
Thanks Ben Menadue for the initial bug report
Refs open-mpi/ompi#6016
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
Though not a recommended configuration it is possible to use Open MPI
over UCX over uGNI. This configuration had some issues related to the
connection management and tl selection. This commit fixes those
issues.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
If UCX is available, then pml/ucx will be used instead of
pml/ob1 + btl/openib, so there is no need to warn about
btl/openib not supporting Infiniband.
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
Add the --pset option for app_contexts so the user can provide a string
name for each app_context. Use the new PMIx pset attribute to store the
names in the PMIx local storage for retrieval
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Now Open MPI requires a C99 compiler. Checking availability of
the following types is no more needed.
- `long long` (`signed` and `unsigned`)
- `long double`
- `float _Complex`
- `double _Complex`
- `long double _Complex`
Furthermore, the `#if HAVE_[TYPE]` style checking is not correct.
Availability of C types is checked by `AC_CHECK_TYPES` in `configure.ac`.
`AC_CHECK_TYPES` defines macro `HAVE_[TYPE]` as `1` in `opal_config.h`
if the `[TYPE]` is available. But it does not define `HAVE_[TYPE]`
(instead of defining as `0`) if it is not available. So even if we
need `HAVE_[TYPE]` checking, it should be `#if defined(HAVE_[TYPE])`.
I didn't remove `AC_CHECK_TYPES` for these types in `configure.ac`
since someone may use `HAVE_[TYPE]` macros somewhere.
Signed-off-by: KAWASHIMA Takahiro <t-kawashima@jp.fujitsu.com>
It seems in some cases (gcc older than v6.0.0) the __atomic_thread_fence is a
no-op with __ATOMIC_ACQUIRE. This appears to be the case with X86_64 so go
ahead and use __ATOMIC_SEQ_CST for the x86_64 read memory barrier. This should
not cause any performance issues as it is equivalent to the memory barrier
in the hand-written atomics.
References #6014
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit added MCA param `opal_max_thread_in_progress` to set the
number of threads allowed to do opal_progress concurrently. The default
value is 1.
Component with multithreaded design can benefit from this change to
parallelize their component progress function.
Signed-off-by: Thananon Patinyasakdikul <tpatinya@utk.edu>
Under certain circumstances, ibv_exp_query_device was
returning an error due to uninitialized fields in the
extended attributes struct.
Fixes: #5810Fixes: #5914
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
In 457f058 I broke the TCP BTL with --enable-ipv6. This patch
fixes the compile error, so IPv6 works again.
Fixed#5996
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
* Needed to properly read PMIx job data like the following
- `OPAL_PMIX_LOCALLDR`
- `OPAL_PMIX_RANK`
- `OPAL_PMIX_GLOBAL_RANK`
- `OPAL_PMIX_APPLDR`
- `OPAL_PMIX_APP_RANK`
Signed-off-by: Joshua Hursey <jhursey@us.ibm.com>
Only default to the external component if its version is
greater or equal than the internal libevent (2.0.22)
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
- Always use the external component when configure'd with --with-libevent=external
- Fix the external libevent library version detection
by testing _EVENT_NUMERIC_VERSION and EVENT__NUMERIC_VERSION macros
- Use the event2/event.h header (event.h is deprecated since libevent 2.0
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
AC_CHECK_DECLS take a comma separated list of macros/symbols,
so replace the whitespace separator with a comma.
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
The monitoring PML hides it's existence from the OMPI infrastructure by
removing itself from the list of PML loaded components, remaining hidden
until MPI_Finalize.
Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
Simplify selection of the address to publish for a given BTL TCP
module in the module exchange code. Rather than looping through
all IP addresses associated with a node, looking for one that
matches the kindex of a module, loop over the modules and
use the address stored in the module structure. This also
happens to be the address that the source will use to bind()
in a connect() call, so this should eliminate any confusion
(read: bugs) when an interface has multiple IPs associated with
it.
Refs #5818
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
Today, a btl tcp module is associated with exactly one IP
address (IPv4 or IPv6). There's no need to reserve space
for both an IPv4 and IPv6 address in the module structure,
since the module will only be associated with one or the
other.
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
Work around a race condition in the TCP BTL's proc setup code.
The Cisco MTT results have been failing on TCP tests due to a
"dropped connection" message some percentage of the time.
Some digging shows that the issue happens in a combination of
multiple NICs and multiple threads. The race is detailed in
https://github.com/open-mpi/ompi/issues/3035#issuecomment-429500032.
This patch doesn't fix the race, but avoids it by forcing
the MPI layer to complete all calls to add_procs across the
entire job before any process leaves MPI_INIT. It also
reduces the scalability of the TCP BTL by increasing start-up
time, but better than hanging.
The long term fix is to do all endpoint setup in the first
call to add_procs for a given remote proc, removing the
race. THis patch is a work around until that patch can
be developed.
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
This commit fixes a deadlock that can occur when using a TL that
supports the connect to endpoint model. The deadlock was occurring
while processing an incoming connection requests. This was done from
an active-message callback. For some unknown reason (at this time)
this callback was sometimes hanging. To avoid the issue the connection
active-message is saved for later processing.
At the same time I cleaned up the connection code to eliminate
duplicate messages when possible.
This commit also fixes some bugs in the active-message send path:
- Correctly set all fragment fields in prepare_src.
- Fix bug when using buffered-send. We were not reading the return
code correctly (which is in bytes). This resulted in a message
getting sent multiple times.
- Don't try to progress sends from the btl_send function when in an
active-message callback. It could lead to deep recursion and an
eventual crash if we get a trace like
send->progress->am_complete->ob1_callback->send->am_complete...
Closes#5820Closes#5821
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
There was a race condition in opal_free_list_get. Code throughout the
Open MPI codebase was assuming that a NULL return from this function
was due to an out-of-memory condition. In some cases this can lead to
a fatal condition (MPI_Irecv and MPI_Isend in pml/ob1 for
example). Before this commit opal_free_list_get_mt looked like this:
```c
static inline opal_free_list_item_t *opal_free_list_get_mt (opal_free_list_t *flist)
{
opal_free_list_item_t *item =
(opal_free_list_item_t*) opal_lifo_pop_atomic (&flist->super);
if (OPAL_UNLIKELY(NULL == item)) {
opal_mutex_lock (&flist->fl_lock);
opal_free_list_grow_st (flist, flist->fl_num_per_alloc);
opal_mutex_unlock (&flist->fl_lock);
item = (opal_free_list_item_t *) opal_lifo_pop_atomic (&flist->super);
}
return item;
}
```
The problem is in a multithreaded environment is *is* possible for the
free list to be grown successfully but the thread calling
opal_free_list_get_mt to be left without an item. The happens if
between the calls to opal_lifo_push_atomic in opal_free_list_grow_st
and the call to opal_lifo_pop_atomic other threads pop all the items
added to the free list.
This commit fixes the issue by ensuring the thread that successfully
grew the free list **always** gets a free list item.
Fixes#2921
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This reverts commit 6acebc40a1.
This patch is causing numerous "Socket closed" messages which are
causing most of the failures on Cisco's MTT run. See
https://github.com/open-mpi/ompi/issues/5849 for more information.
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
Looks like a filename was missed when pmix sucked in the installdirs
framework. Fixing the typo fixes "make ctags" and "make cscope".
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
It is apparently possible for different instances of the same UCT
transport to have different limits (max short put for example). To
account for this we need to store the attributes per TL context not
per TL. This commit fixes the issue.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
While trying to debug #3035, it's not clear whether there is
an issue with the modex data or printing the address list.
Print the number of endpoints on the error, which will help
determine which case is happening to Cisco.
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
When creating TCP BTL modules, print more information about the
module's ethernet association, including the first address associated
with the device, as debug output.
Fix a flipped output string for IPv4 and IPv6 addresses in the
modex send code.
Add the addresses being published in the modex to the debugging
output in modex send, to help match failures in endpoint match.
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
This commit updates the uct btl to change the transports parameter
into a priority list. The dc_mlx5, rc_mlx5, and ud transports to the
priority list. This will give better out of the box performance for
multi-threaded codes beacuse the *_mlx5 transports can avoid the mlx5
lock inside libmlx5_rdmav2.
This commit also fixes a number of leaks and a possible deadlock when
using RDMA.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
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>
Open MPI's developers like to assume that asprintf() always sets
the ptr to NULL on error, but the standard (and Linux glibc) do
not guarantee this. As a result, we're making opal_asprintf()
always available for developers, which will guarantee that
ptr is set to NULL on error.
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
If we are using the internal PMIx component and the embedded library fails to configure, then fail - don't silently fail to build and then fail in execution
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
We only need to pass a custom range if the target is a single process.
Otherwise, we let the range be "session".
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
This commit works around an Oracle C compiler bug in 5.15 (not sure
when it was introduced). The bug is triggered when we chain
assignments of atomic variables. Ex:
_Atomic intptr x, y;
intptr_t z = 0;
x = y = z;
Will produce a compiler error of the form:
operand cannot have void type: op "="
assignment type mismatch:
long "=" void
To work around the issue we are removing the chain assignment and
setting the head and tail on different lines.
Fixes#5814
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
On some platfoms reading a 64-bit value is non-atomic and it is
possible that the two 32-bit values are read in the wrong order. To
ensure the tag is always read first this commit reads the tag before
reading the full 64-bit value.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Remove the pack/unpack pragma around net/if.h on MacOS, which
was added to fix a bug in MacOS X 10.4.x on 64-bit platforms.
The bug was fixed in Mac OS X 10.5.0 and, sometime in the last
11 years, compilers started emitting warnings about the fact
that the Apple header stomped over the pragma pack settings
from the workaround. We already don't support versions of MacOS
earlier than 10.5, so there's no point in keeping the workaround.
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
Open MPI doesn't support any transports on MacOS which require
memory manager hooks. The memory patcher component uses the
syscall interface, which has been deprecated in recent versions
of MacOS. Since we don't need it and it emits warnings about
deprecation, disable the memory patcher component on MacOS.
Fixes#5671
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
Get Brian's patch from #5825 and his log message:
Fix a failure in binding the initiating side of a connection
on MacOS. MacOS doesn't like passing the size of the storage
structure (sockaddr_storage) instead of the expected size of
the structure (sockaddr_in or sockaddr_in6), which was causing
bind() failures. This patch simply changes the structure size
to the expected size.
Add a more clear error message in debug mode.
Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
Per
https://github.com/open-mpi/ompi/issues/3035#issuecomment-426085673,
it looks like the IP address for a given interface is being stashed in
two places: on the endpoint and on the module.
1. On the endpoint, it is storing the moral equivalent of a
(struct sockaddr_in.sin_addr).
2. On the module, it is storing a full (struct sockaddr_storage).
The call to opal_net_get_hostname() expects a full (struct sockaddr*)
-- not just the stripped-down (struct sockaddr_in.sin_addr). Hence,
when the original code was passing in the endpoint's (struct
sockaddr_in.sin_addr) and opal_net_get_hostname() was treating it
like a (struct sockaddr), hilarity ensued (i.e., we got the wrong
output).
This commit eliminates the call to opal_net_get_hostname() and just
calls inet_ntop() directly to convert the (struct
sockaddr_in.sin_addr) to a string.
NOTE: Per the github comment cited above, there can be a disparity
between the IP address cached on the endpoint vs. the IP address
cached on the module. This only happens with interfaces that have
more than one IP address. This commit does not fix that issue.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Add a hueristic: if the string copy is "too long", fail an assert().
This is based on the premise that Open MPI doesn't do large string
copies. So if we see a dest_len that is over a certain threshhold
(currently set at 128K), this is likely a programmer error, and on
debug builds, we should fail an assert(). In production builds, it
will work just fine (assuming that it's not a programmer error).
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
In many cases, this was a simple string replace. In a few places, it
entailed:
1. Updating some comments and removing now-redundant foo[size-1]='\0'
statements.
2. Updating passing (size-1) to (size) (because opal_string_copy()
wants the entire destination buffer length).
This commit actually fixes a bunch of potential (yet quite unlikely)
bugs where we could have ended up with non-null-terminated strings.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Do essentially the same thing as strncpy(3), but a) ensure to always
terminate the destination buffer with a \0, and b) do not \0-pad to
the right.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Fix the test that determined whether we output "writeable" or
"read-only" for MCA vars (it was checking the wrong flag).
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Disable async receive for CUDA under OpenIB. While a performance
optimization, it also causes incorrect results for transfers
larger than the GPUDirect RDMA limit. This change has been validated
and approved by Akshay.
References #3972
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
This can be returned when running on QEMU user-mode emulation,
which does not support getsockopt with SO_RCVTIMEO.
Signed-off-by: Michael Kuron <mkuron@icp.uni-stuttgart.de>
According to clang on MacOS, passing a bool parameter -- which
undergoes default parameter promotion -- to va_start() results in
undefined behavior. So just change these params to int and avoid the
issue.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
This commit disables the use of both the builtin and hand-written
atomics if proper C11 atomic support is detected. This is the first
step towards requiring the availability of C11 atomics for the C
compiler used to build Open MPI.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit updates the entire codebase to use specific opal types for
all atomic variables. This is a change from the prior atomic support
which required the use of the volatile keyword. This is the first step
towards implementing support for C11 atomics as that interface
requires the use of types declared with the _Atomic keyword.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
To ensure fast box entries are complete when processed by the
receiving process the tag must be written last. This includes a zero
header for the next fast box entry (in some cases). This commit fixes
two instances where the tag was written too early. In one case, on
32-bit systems it is possible for the tag part of the header to be
written before the size. The second instance is an ordering issue. The
zero header was being written after the fastbox header.
Fixes#5375, #5638
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Since openib is on its long, slow way out the door, don't let it
complain about not being able to find any NICs at run time.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
This commit updates the patcher component to either use the
__clear_cache intrinsic or the correct assembly to flush the
instruction cache.
Fixes#5631
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes a bug when using the UCT btl with the UCX memory
hooks disabled. We were misssing a call to
opal_mem_hooks_unregister_release to remove the btl memory hook
callback.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Ensure that the project, framework, component, and variable names are
lower than max lengths. This is a follow-on to
992a8e8297, per discussion on
https://github.com/open-mpi/ompi/pull/5642.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
If someone specifies --with-verbs-usnic, actually do a configury check
to ensure that it will compile (vs. assuming that it will compile if
someone asks for it).
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Since version hwloc 2.0.0 has a new organization of NUMA nodes on the
topology tree. This commit adds the detection of local NUMA object for
hwloc => 2.0.0, which fixes the procs bindings policy for rmaps mindist
component.
Signed-off-by: Boris Karasev <karasev.b@gmail.com>
bugfix: major: openib send credits returned correctly after a fault for pending frags to dead processes; also tweak the default IB retry timeouts tomake this happen faster
Make it compile in non-debug builds
Mark the IB endpoint as failed when invoking an error; this resolves UDCM connection deadlocks
Changing the default IB retry timeouts is not a good idea.
We'll need to find another way to speedup credit recovery in failure cases.
Remove ULFM specific cases
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
The important part of this fix is a couple places 5 was hard-coded that needed to be
strlen(OPAL_INFO_SAVE_PREFIX).
But also this contains a fix for a gcc 7.3.0 compiler warning about snprintf(). There
was an "if" statement making sure all the arguments had appropriate strlen(), but gcc
still complained about the following snprintf() because the size of the struct element
is iterator->ie_key[OPAL_MAX_INFO_KEY + 1].
Signed-off-by: Mark Allen <markalle@us.ibm.com>
When an error is returned by the socket operations, trigger the
appropriate error path in the PML to give an opportunity for
rerouting/error handling.
Signed-off-by: Aurelien Bouteiller <bouteill@icl.utk.edu>
The write memory barrier was intended to precede setting a fast-box
header but instead follows it. This commit moves the memory barrier to
the intended location.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
The Autoconf AC_CONFIG_* macros can only be instantiated exacly once
for any given file, *and* they must be in a code execution path at run
time for the target file to be generated at the end of configure.
For example, if you want to generate file ABC at the end of configure,
you must invoke the AC_CONFIG_FILES(ABC) macro in a code path that
will get executed when configure is run.
That's pretty straightforward.
What's not straightforward is two corner cases:
1. You cannot invoke the AC_CONFIG_FILES(ABC) macro for the same file
more than once. If you do, autoreconf will fail (even before you
can run configure).
2. If AC_CONFIG_FILES(ABC) is not in a code path that is executed by
configure, the file ABC is not registered properly, and ABC will
not be generated at the end of configure.
This applies to hwloc because hwloc's HWLOC_SETUP_CORE macro calls
both AC_CONFIG_FILES and AC_CONFIG_HEADER to setup its Makefiles
(etc.) so that targets like "make distclean" and "make distcheck" will
work properly. Hence, we *have* to invoke HWLOC_SETUP_CORE.
However, the MCA_opal_hwloc_hwloc201_CONFIG macro has a few side
effects. It would be nice to do able to do something like this:
```
if hwloc:extern is going to be used:
Invoke minimal HWLOC_SETUP_CORE (with no side effects)
else
Invoke full HWLOC_SETUP_CORE (with side effects)
fi
```
But we can't, because autoreconf will detect that AC_CONFIG_FILES has
been invoked on the same files more than once (regardless of whether
those code paths will be executed at run time or not). Kaboom.
Similarly, we can't do this:
```
if hwloc:extern is not going to be used:
Invoke full HWLOC_SETUP_CORE (with side effects)
fi
```
Because then hwloc's AC_CONFIG_FILES won't be registered properly when
hwloc:external *is* used (i.e., when the HWLOC_SETUP_CORE macro is not
in a code path that is executed at run time), and targets like "make
distclean" will fail because hwloc's Makefiles won't have been setup.
Kaboom.
But remember that the hwloc framework is a bit special: there will
only ever be 2 comoponents: external and internal. External is
guaranteed to be configured first because of its priority. So the
internal component (i.e., this component) immediately knows if it is
going to be used or not based on whether the external component
configuration succeeded or failed.
Specifically: regardless of whether the internal component (i.e., this
component) is going to be used, we have to invoke HWLOC_SETUP_CORE.
But we can manage the side effects: allow the side effects when
this/internal component is going to be used, and avoid the side
effects when this/internal component is not going to be used.
This is a little less clean than I would have liked, but because of
Autoconf's oddity about its AC_CONFIG_* macros, this is the only
solution I could come up with.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
In order to make "make distclean" (and friends) work, we need to
*always* invoke the embedded configure script -- even if we know that
we're not going to use this component.
But in cases where we know we're not going to use this component, we
also need to avoid the side effects of the code path that is used when
we *do* want to use this component. So split the two possibilities
into two different macros:
1. MCA_opal_event_libevent2022_FAKE_CONFIG: which does almost nothing
except invoke the underlying "configure" script.
2. MCA_opal_event_libevent2022_REAL_CONFIG: which does all the real
work (including invoking the underlying "configure" script).
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
We know that event:external will be configured first (because of its
priority). Take advantage of that here in libevent2022 by having it
refuse to configure / politely fail if event:external succeeded.
Also print out some additional lines in configure output indicating
what is going on (i.e., event:external succeeded, so this component
will be skipped, or event:external failed, so this component will be
used).
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
We know that hwloc:external will be configured first (because of its
priority). Take advantage of that here in hwloc201 by having it
refuse to configure / politely fail if hwloc:external succeeded.
Also print out some additional lines in configure output indicating
what is going on (i.e., hwloc:external succeeded, so this component
will be skipped, or hwloc:external failed, so this component will be
used).
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
libevent does not support multiple threads calling the event loop on
the same event base. This causes external libevent's to print out
re-entrant warning messages. This commit fixes the issue by protecting
the call to the event loop with an atomic swap check.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
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>
Things got a little out of whack and we weren't actually processing the map-by modifiers, plus an error crept into the display of the binding report. So clean those up.
Thanks to @tonyreina for the error report
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Per https://github.com/open-mpi/ompi/issues/5031, if the user didn't specify a particular PMIx installation, then default back to the internal version if it is newer than the discovered external one. PMIx doesn't yet provide a full signature so we have to just get as close as possible for now.
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Due to decreasing support by vendors/other orgs for the OpenIB BTL,
only look for iWarp/RoCE devices by default. Allow IB HCAs
with ports configured for ethernet.
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
OFI BTL uses context for completion but never ask for it in
fi_getinfo(3). This commit makes sure that we always ask for FI_CONTEXT
to eliminate any potential error.
Signed-off-by: Thananon Patinyasakdikul <thananon.patinyasakdikul@intel.com>
This commit fixes two bugs in the RMA/atomic emulation code:
1) Fix a fragment leak when using AMO emulation.
2) Always initialize the single-copy emulation code. This is required
to use the AMO support.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit updates the atomic fifo code to fix a consistency issue
observed on Power9 systems when builtin atomics are used. The cause
was two things: 1) a missing write memory barrier in fifo push, and 2)
a read ordering issue when reading the fifo head non-atomically. This
commit fixes both issues and appears to correct then inconsistency.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
If OMPI is configured with `--with-hwloc=external` or `--with-hwloc=DIR`
and gfortran is used, I see a lot of warnings when compiling files
under the `ompi/mpi/fortran` directory.
```
f951: Warning: Nonexistent include directory
'BUILD_DIR/opal/mca/hwloc/external/hwloc/include' [-Wmissing-include-dirs]
```
There is no such `include` directory in the source tree and `configure`-
created tree. I think these lines in the `configure.m4` file are wrongly
copied from that for the embedded `hwlocXXX` component in the past.
The `-Wmissing-include-dirs` option is enabled in gfortran by default
but it is not enabled by default (or even with `-Wall`) in gcc and g++.
Signed-off-by: KAWASHIMA Takahiro <t-kawashima@jp.fujitsu.com>
- added common logging infrastructure for all
UCX modules
- all UCX modules are switched to new infra
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
The descriptor flags field in a fragment were being ready after the
fragment may have been freed. This commit reads the flags before
calling the user callback.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit adds support for atomic operations as well as rdma for
systems without rdma support. This support is implemented using an
internal send tag.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
- some common functionality of del_procs calls is moved into
mca_common module
- blocking ucp_put call is replaced by non-blocking routine
Signed-off-by: Sergey Oblomov <sergeyo@mellanox.com>
The current cast is *functional*, but isn't really the way it should
be done. This commit makes the cast the way it should be done.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Fix two facepalms:
1. The "uint32" in the hash map functions refer to the *key* size, not
the *value* size. The values are always 64 bits.
2. Pass the straight value to the "set" functions -- not the pointer
to the value.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
This commit changed the way btl/ofi call progress. Before, we force
progression with every rdma/atomic call. This gives performance boost in
some case and slow down on others. Now we only force progression after
some number of rdma calls which result in better performance overall.
Also added new MCA parameter 'mca_btl_ofi_progress_threshold' to set
the threshold number. The new default is 64.
Also:
Added FI_DELIVERY_COMPLETE to tx_rtx flags to ensure that the completion
is generated after the message has been received on the remote side.
Signed-off-by: Thananon Patinyasakdikul <thananon.patinyasakdikul@intel.com>
This commit improves the injection rate and latency for RDMA
operations. This is done by the following improvements:
- If C11's _Thread_local keyword is available then always use the
same virtual device index for the same thread when using RDMA. If
the keyword is not available then attempt to use any device that
isn't already in use. The binding support is enabled by default but
can be disabled via the btl_ugni_bind_devices MCA variable.
- When posting FMA and RDMA operations always attempt to reap
completions after posting the operation. This allows us to
better balance the work of reaping completions across all
application threads.
- Limit the total number of outstanding BTE transactions. This
fixes a performance bug when using many threads.
- Split out RDMA and local SMSG completion queue sizes. The RDMA
queue size is better tuned for performance with RMA-MT.
- Split out put and get FMA limits. The old btl_ugni_fma_limit MCA
variable is deprecated. The new variable names are:
btl_ugni_fma_put_limit and btl_ugni_fma_get_limit.
- Change how post descriptors are handled. They are no longer
allocated seperately from the RDMA endpoints.
- Some cleanup to move error code out of the critical path.
- Disable the FMA sharing flag on the CDM when we detect that there
should be enough FMA descriptors for the number of virtual devices
we plan will create. If the user sets this flag we will not unset
it. This change should improve the small-message RMA performance by
~ 10%.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit adds a new btl for one-sided and two-sided. This btl
uses the uct layer in OpenUCX. This btl makes use of multiple uct
contexts and per-thread device pinning to provide good performance
when using threads and osc/rdma. This btl has been tested extensively
with osc/rdma and passes all MTT tests on aries and IB hardware.
For now this new component disables itself but can be enabled by
setting the btl_ucx_transports MCA variable with a comma-delimited
list of supported memory domains/transport layers. For example:
--mca btl_uct_memory_domains ib/mlx5_0. The specific transports used
can be selected using --mca btl_uct_transports. The default is to use
any available transport.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
The giant size of the TCP proc struct is causing a problem in some
environments (because it is allocated on the stack), and it was too
big, anyway.
Instead, use a hash map. That way, it starts small and can grow if it
needs to. It also makes no assumptions about the values of the kernel
interface indexes.
Fixes#5292.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Since the new binding option is tied to the --cpu-list orterun CLI
option, make the --bind-to option reflect the same name (vs. the
--cpu-set CLI option, which is entirely different). For example:
mpirun --bind-to cpu-list:ordered ...
Note that "--bind-to cpulist:ordered" is accepted as a synonym,
because people will be lazy.
Also add some minor updates to the orterun.1in man page for
clarification.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
