[mpich-commits] [mpich] MPICH primary repository annotated tag, v3.1rc3, created. v3.1rc3

mysql vizuser noreply at mpich.org
Tue Feb 4 17:20:12 CST 2014


This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "MPICH primary repository".

The annotated tag, v3.1rc3 has been created
        at  c0fe4f463eed85052d58bd131b51f9ce4d71e5d0 (tag)
   tagging  1bef988953257fb6acababd5e3af17bad2b91e9c (commit)
  replaces  v3.1rc2
 tagged by  Ken Raffenetti
        on  Tue Feb 4 13:20:32 2014 -0600

- Log -----------------------------------------------------------------
tagging 'v3.1rc3'

Antonio J. Pena (12):
      Removed ancient comments in dims_create.c
      Fixed suboptimal distributions in MPI_Dims_create
      Added regression test for ticket #1765
      Fix warnings in ch3u_rma_acc_ops and ch3u_rma_ops
      Fix compiler warning in socksm.c
      Fix warnings in initcxx.cxx with Portland compiler
      Fix warnings in initcxx.cxx with Portland compiler
      Fix compiler warnings about unused variables
      Fix compiler warnings about variables for assert
      Enclose into #ifdef code causing compiler warnings
      Fix warnings in variables only used in assertions
      Fix compiler warnings in hwloc with --enable-fast

Bob Cernohous (1):
      Test updates for ad_bglockless.

Gilles Gouaillardet (3):
      Use MPI_Count with MPIR_Status_set_bytes.
      Use MPI_Type_size_x.
      Fix prototypes and remove unused local variables.

Haizhu Liu (3):
      Exit with error when dynamic tasking is used in an invalid environment
      PAMID support of request based rma functions
      PAMID support of rma atomic functions

Huiwei Lu (1):
      Updates testing scripts to generate junit output

James Dinan (2):
      Improve context ID exhaustion error reporting
      Update comm_create to use sparse ctx id allocation

Junchao Zhang (34):
      Fix a typo about MPIR_CVAR_NEM_*
      Make extracterrmsgs know MPI_T error codes
      Add MPIR_ERRTEST_* macros for MPI_T
      Improve MPI_T error handling
      Revise the test to add MPI_T_PVAR_ALL_HANDLES
      Delete mpiinstr.h since it is superceeded by MPI_T
      Add rma to option --enable-mpit-pvars=list
      Change prototype of MPIR_T_pvar_get_value_cb etc
      Let MPIR_T_PVAR_*_GET_VAR() act as a rvalue
      Add a counter into structure MPIR_T_pvar_timer_t
      Add macro MPIR_T_PVAR_{COUNTER, TIMER}_ADDR
      Fix a bug in MPI_T watermark pvar registeration
      Port instr variables in rma code to MPI_T pvars
      Use if (ENABLE_PVAR_NEM) instead of #ifdef
      Enclose macro arguments in parentheses
      Delete cvars.yml since it will be split.
      Rename script gencvars to extractcvars
      Implement extractcvars, which extracts cvar info.
      Fix a wrong cvar name
      Revert commented out code, to make a cvar work
      Spread out cvar info into source files
      Revise description of MPIR_CVAR_BCAST_* cvars
      Add the missing NEMESIS_LMT_DMA_THRESHOLD
      Add indentions, remove white spaces, replace Tabs
      Make ReadInterface extract both arg types & names
      Add indentions, remove white spaces, replace Tabs
      Print dummy arg names defined by MPI to f90 files
      Fix a script problem appearing in old perl
      Add prefix MPI_CVAR_ to a cvar name in description
      white space fixup
      Add "See also" in descriptions of related cvars
      Disable Clang's type checking for MPI_2INT
      Add return 0 to the main function of tests
      Add -lgcov to LIBS when --enable-coverage

Ken Raffenetti (32):
      require 8-byte memory address for large mem tests
      fixup for 1638ed88
      correct alias option test for weak attributes
      Revert "correct alias option test for weak attributes"
      strengthen test for weak alias attribute
      set context mask depending on need for ICOMM_WORLD
      disable win_large_shm test when unsupported
      extend timeout for large sendrec test
      warning squashes for hwloc
      format warning squash
      patch to detect bgq compilers in config.rpath
      corrections to weak alias test
      Improve pkg-config support
      compile wrapper cleanup
      gitignore generated mpivars file
      corrections to compile wrapper changes
      simplify bitmap intialization in process binding
      example program to display processor affinity
      hydra process binding/mapping overhaul
      remove help text for unsupported mapping option
      add debugging for process affinity options
      make topology debugging output more reliable
      add manual test for process affinity
      mark cas_type_check test as xfail
      handle when qsort is not available in vc setup
      use pmi as default name service
      fix for unpublish name behavior
      error reporting corrections in pmi nameservice
      corrects the MPI error class returned from MPI_Unpublish_name.
      check for correct errclass in unpub test
      return correct error code in MPI_UNPUBLISH_NAME
      remove unnecessary system-out data in junit

Masamichi Takagi (23):
      Add netmod files for Xeon Phi on DCFA/McKernel
      Use MPIR_STATUS_SET_COUNT for the request struct
      Comment out debug messages in dcfa_impl.h
      Correct error string
      Add Fortran related cross compilation values
      Modify macro name for DCFA specific lines
      Enable --with-dcfa* configure options
      Add a description on the DCFA network module
      Auto-detect IB libraries for the DCFA netmod
      Eliminate --enable-strict warnings and errors
      Increase max number of connections
      Correct reference overrun in debug message
      Modify credit information packet rate
      Correct debug messages
      Remove macro for receiver-initiate lmt
      Rename the DCFA netmod to the IB netmod
      Add a header file to Makefile.mk of ib netmod
      Replace MPIU_QUOTE with MPIDI_QUOTE
      Remove obsolete code
      Correct references to "subtype" of netmod packets
      Modify README.vin on description of the IB netmod
      Fix type names in the IB netmod
      Fix warnings and errors with --enable-strict

Michael Blocksome (6):
      PAMId: added some mpi 3.0 RMA function support
      Set the maximum number of PAMI contexts to use to '1' for PE
      Add missing FCNAME #define for error reporting.
      fix missing const for MPID_Accumulate and MPID_Put functions.
      gcc 4.4.6 warning cleanup.
      bgq fix for mpivars link error

Nysal Jan K.A (2):
      Initialize the dbg_next field of the mpi request
      Fix MPI_Attr_put on big-endian systems

Pavan Balaji (54):
      Move brace outside of the ifdef, so it's always defined.
      Remove misplaced brace.
      White-space cleanup.
      Check for number of arguments instead of a NULL argument.
      Upgrade to hwloc-1.8
      Hydra-specific patches for hwloc.
      Update CHANGES file for hwloc-1.8.
      Added more MPICH authors and removed old ones.
      White space cleanup.
      White-space cleanup for redability.
      Added a new comm_idup_overlap test.
      Create a dummy stdin socket instead of simply closing it.
      Move default eager threshold setting to ch3.
      Add primes to cover till 2 billion MPI ranks.
      Added a new dims_create test.
      Change variable name and add more comments.
      Bug-fix: reset the rmk when the hostlist came from the user.
      Bug-fix: win_req was never initialized.
      Remove duplicate code.
      Make SMP-awareness in MPI_Barrier consistent.
      pamid is capable of supporting WIN_UNIFIED.
      Thread critical-section initialization fixes.
      We always need the runtime thread-safety check.
      Prioritize thread-multiple branch.
      Comment out nb_test, since it's not entirely correct.
      White space cleanup.
      Time iterations and break out if we are too slow.
      Rename MPICH2 to MPICH.
      Update the MPICH ABI string for the 3.1rc3 release.
      Use MPIU_Malloc/Free instead of malloc/free.
      White space cleanup.
      Move communicator destruction to after progress checks.
      Revert "Move communicator destruction to after progress checks."
      Remove unused flag.
      Improve PMI_process_mapping parsing.
      Bug-fix: deal with wrap around PMI process mapping strings
      Move communicator destruction to after progress checks.
      Provide a way to disable ssh warnings in Hydra.
      Remove xfail for ticket-1996, which is closed.
      Delete some windows-related code.
      Fix broken release script from [ba3badff].
      Require correct autotool versions for a release.
      Update copyright information for Jeff Hammond's tests.
      Added a bunch of other copyrights.
      Revamp how enable-fast is used.
      Update README to tell users how to do performance testing.
      Remove MPICH_FAST usage from the test suite.
      Bug-fix: bad declaration when error checking is disabled.
      Reduce the number of processes needed by rma-contig.
      Rename binding reference files to be windows-friendly.
      Update CHANGES file for the rc3 release.
      Remove release notes that are no longer relevant.
      Remove .bat files.
      Remove .bat file checks in Makefile.

Rob Latham (20):
      fix compile errors with lustre
      ROMIO standalone: check for MPI_Count
      ROMIO standalone: remove internal headers
      ROMIO standalone: check HINDEXED_BLOCK
      ROMIO standalone: check for need of const
      deal with "infinitely stale" case
      re-enable deferred open on Blue Gene
      Introduce and use ADIO_TWO_PHASE feature-check
      deferred open: fixup immediate-mode communicators
      unused variable warning
      ROMIO standalone: actually install library
      ROMIO standalone: fix up link-time errors
      ROMIO standalone: check for large-count routines
      bluegene open fixup: don't broadcast garbage
      bring back include file needed for MPI_AINT_CAST_TO_VOID_PTR
      Revert "bring back include file needed for MPI_AINT_CAST_TO_VOID_PTR"
      Use ROMIO pointer-casting routines
      release memory in error path
      a partial round of datatype optimizations
      Revert "a partial round of datatype optimizations"

Sameer Kumar (2):
      Implement MPID_Get_accumulate()
      Bug fix for strided datatypes.

Sameh Sharkawi (6):
      Task flooding causes poor MPI_Reduce performance
      Fixes for Collective Selection enablement
      Fixes for MPI_Alltoall when using non-contig data and collective selection
      Debug lib memory tracing signature
      Allow env variables to control memory initialization.
      PAMID fix for schedule release

Su Huang (9):
      PAMId: MPI_Win_set_info, MPI_Win_get_info and the frame work for window creation
      PAMID: Support MPI_PROC_NULL in MPI_Win_lock and MPI_Win_unlock
      PAMID: update PE trace code to support MPI_ANY_SOURCE
      PAMID: MPID_Win_allocate_shared and MPID_Win_shared_query support
      PAMID: fixes for MPI_Win_*
      PAMID: segfault occurred in MPIDI_Win_DoneCB
      Fix multiple FCNAME declarations error.
      PAMID: fix memory leak problem
      PAMID:fix memory leak and BGQ problems

Wesley Bland (9):
      Update the changes file for 3.1rc2
      Cleanup generated test files.
      Add doctext to hydra build system.
      Add man page for missing hydra executables
      Mark the ft/die test as xfail for now.
      Mark ft/sendalive as xfail for now.
      Moves the tag reservation to MPI layer
      Remove a comment that doesn't apply anymore.
      Add configure option to disable FT tests

William Gropp (7):
      Minor typo corrections in PAC macros comments
      Fix check for datatype name
      Add program to show MPI_T variables
      Fix typo in library choice for mpivars
      Add mpivars to build
      Fix typos in mpivars
      Fix bogus datatype perf test

-----------------------------------------------------------------------


hooks/post-receive
-- 
MPICH primary repository


More information about the commits mailing list