Prep for 5.0.9amzn1 release #13530
Merged
Prep for 5.0.9amzn1 release #13530
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
`requested` in `MPI_Init_thread` would invoke the error handler, even though it is an useful override in some threaded library use cases. Signed-off-by: Aurelien Bouteiller <abouteil@amd.com> (cherry picked from commit 27332fc)
(single,etc) in addition to numeric 0-3 values Signed-off-by: Aurelien Bouteiller <abouteil@amd.com> (cherry picked from commit 3de2489)
…ages
Including, but not limited to:
* Added much more description of and distinction between the MPI world
model and the MPI session model. Updated a lot of old,
pre-MPI-world-model/pre-MPI-session-model text that was now stale /
outdated, especially in the following pages:
* MPI_Init(3), MPI_Init_thread(3)
* MPI_Initialized(3)
* MPI_Finalize(3)
* MPI_Finalized(3)
* MPI_Session_init(3)
* MPI_Session_finalize(3)
* Numerous formatting updates
* Slightly improve the C code examples
* Describe the mathematical relationship between the various
MPI_THREAD_* constants in MPI_Init_thread(3)
* Note that the mathematical relationships render nicely in HTML,
but don't render entirely properly in nroff. This commit author
is of the opinion that the nroff rendering is currently "good
enough", and some Sphinx maintainer will fix it someday.
* Add descriptions about the $OMPI_MPI_THREAD_LEVEL env variable and
how it is used in MPI_Init_thread(3)
* Added more seealso links
Signed-off-by: Jeff Squyres <jeff@squyres.com>
(cherry picked from commit aff3afd)
…it doc. Signed-off-by: Aurelien Bouteiller <abouteil@amd.com>
Signed-off-by: Aurelien Bouteiller <abouteil@amd.com>
Thanks to Ben Menadue for pointing out that ompi_fortran_string_c2f() missed a case to properly terminate the resulting Fortran string when copying from a longer C source string. Signed-off-by: Jeff Squyres <jeff@squyres.com> (cherry picked from commit 694e78a)
The table added in 061f908 (A variety of docs updates:, 2022-09-12) mentioning the different prefixes for Open MPI, PMIx and PRRTE MCA parameters set via environment variables has one too many "R"'s in 'PRRTE_MCA_': the correct prefix is 'PRTE_MCA_'. Fix that, and make it clear that it is not a typo. Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com> (cherry picked from commit bd9adb4)
…-c2f-string-copy v5.0.x: fortran: fix ompi string c2f where len(fstr) < len(cstr)
…ar-prefix-5.0 v5.0.x: docs/mca.rst: fix MCA environment variable prefix for PRRTE
…level-ignored@v5 v5.0.x: Thread level set from ENV crashes (cherry open-mpi#13211)
PMIx v5.0.9 PRRTE v3.0.12 Signed-off-by: Ralph Castain <rhc@pmix.org>
Check PMIx/PRRTE release branches prior to release
… or failure The MCA_PML_OB1_ADD_ACK_TO_PENDING method creates a mca_pml_ob1_pckt_pending_t to hold an ack to be sent later. This method builds the pending packet then puts it on the mca_pml_ob1.pckt_pending list for later transmission. It does not, however, set the required hdr_size field on the struct. This leads to issues when the packet is later sent because it could contain any value. With some btls this will lead to memory corruption (if the size is not checked against btl_max_send_size) or just allocation failure because the size is too big. In other situations it could lead to a truncated packet being send (if the size previously in hdr_size is smaller than an ack). To fix the issue this commit gets rid of the macro entirely and replaces it with a new inline helper method that does the same thing. This helper uses the existing mca_pml_ob1_add_to_pending helper (which sets hdr_size) to reduce duplicated code. Tested and verified this fixes a critical issue triggered on our hardware. Signed-off-by: Nathan Hjelm <hjelmn@google.com> (cherry picked from commit 48490b9)
…where_pending_packets_can_have_incorrect_header_sizes Fix bug in MCA_PML_OB1_ADD_ACK_TO_PENDING that causes memory overruns…
- Update VERSION file to v5.0.9rc1 with correct date (23 September 2025) - Update NEWS with actual changes from v5.0.8 to v5.0.9rc1 including: * PMIx v5.0.9 and PRRTE v3.0.12 updates * GPFS 5.2.3-0+ support * OFI accelerator memory enhancements * Critical PML OB1 bug fix for memory overruns * Fortran string conversion fixes * Threading improvements * Various documentation and build system fixes Signed-off-by: Tomislav Janjusic <tomislavj@nvidia.com>
Signed-off-by: Mikhail Brinskii <mikhailb@nvidia.com>
Signed-off-by: Sergey Lebedev <sergeyle@nvidia.com> (cherry picked from commit 0caae60)
V5.0.x: v5.0.9rc1
…ocal_id_v5 COLL/UCC: set node local id - v5.0.x
OMPI/MCA/PML/UCX: Set node local id - v5.0.x
In some cases the CUDA install directory contains two libcuda.so and this breaks OMPI CUDA detection. Pick the first of these libraries seems to be a good soltuion for all cases. Signed-off-by: George Bosilca <gbosilca@nvidia.com> (cherry picked from commit c7e27b9)
Signed-off-by: xbw <78337767+xbw22109@users.noreply.github.com> (cherry picked from commit 0999325)
v5.0.x: Only pick one CUDA
Signed-off-by: charlesgwaldman <120225331+charlesgwaldman@users.noreply.github.com> (cherry picked from commit a6b8cd3)
Fix `see-also` errors in the document (v5.0.x)
Use unique, NVIDIA-specific workflow names so that it's easier to identify these workflows on the github dashboard backend. Signed-off-by: Jeff Squyres <jeff@squyres.com> (cherry picked from commit dcac103)
…ia-github-actions NVIDIA github workflows: use unique workflow names (v5.0.x)
…ng-fix Update history.rst (spelling) (v5.0.x)
Signed-off-by: Tomislav Janjusic <tomislavj@nvidia.com>
v5.0.x: prepare v5.0.9rc2 release
Signed-off-by: Kento Hasegawa <hasegawa.kento@fujitsu.com> (cherry picked from commit 4b1b9a9)
…nitialization COLL/UCC: Fix initialization in non-blocking (v5.0.x)
Signed-off-by: Tomislav Janjusic <tomislavj@nvidia.com>
v5.0.9: consolidate news.
Open MPI v5.0.9
Signed-off-by: Jessie Yang <jiaxiyan@amazon.com> (cherry picked from commit f65f900)
Add FI_COMPLETION flag to ensure completion entries are generated for all data transfer operations. Signed-off-by: Jessie Yang <jiaxiyan@amazon.com> (cherry picked from commit 15fe246)
Share the domain between the MTL and BTL layers to reduce the total number of domains created. This helps avoid hitting system resource limits on platforms with high core counts. Instead of having the common code allocate a single domain with the superset of all required capabilities, we attempt to reuse an existing fabric and domain if the providers can support MTL’s and BTL’s different capability sets. This approach allows providers that support domain sharing to reuse resources efficiently while still preserving flexibility. If the providers cannot reuse the fabric and domain due to incompatible requirements, separate domains will be created as before. Signed-off-by: Jessie Yang <jiaxiyan@amazon.com> (cherry picked from commit 69d2737)
Signed-off-by: Brian Barrett <bbarrett@amazon.com>
jiaxiyan
approved these changes
Nov 24, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
13 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Release 5.0.9 + the single Fabric/Domain per process patch series (which will be in v5.0.10).
bot:notacherrypick