summaryrefslogtreecommitdiffstats
path: root/src/H5FDmpio.c
Commit message (Collapse)AuthorAgeFilesLines
* Warnings fixes (#1680)jhendersonHDF2022-04-251-1/+1
|
* Retrieve MPI-IO hints used by MPI library after file open (#1636)Wei-keng Liao2022-04-151-0/+49
| | | | | | H5Pget_fapl_mpio() should return an MPI info object containing all the MPI-IO hints used by the MPI library underneath, after the file is opened. Some hints, such as cb_nodes (number of I/O aggregators), are useful for HDF5 applications and I/O libraries built on top of HDF5.
* Removes the STATIC flavor of FUNC_ENTER macros (#1622)Dana Robinson2022-04-081-17/+17
| | | | | * Removes the STATIC flavor of FUNC_ENTER macros
* Add H5_CHECK_MPI_VERSION macro to check for at least MPI version of ↵jhendersonHDF2022-03-301-7/+7
| | | | | | | | | (version, subversion) (#1556) * Fix MPI version check for MPI_Aint_diff usage * Add H5_CHECK_MPI_VERSION macro Adds new H5_CHECK_MPI_VERSION macro to check for MPI library version/subversion pair of at least the specified pair
* Merge initial version of selection I/O feature into develop (#1367)Neil Fortner2022-03-261-37/+1098
|
* Parallel Compression improvements (#1302)jhendersonHDF2022-02-241-5/+39
|
* Implement support for big I/O for independent reads (#1446)Neil Fortner2022-02-181-16/+34
| | | | | | | | | | * Implement big I/O support for independent reads. Add test for this. * Committing clang-format changes * Update big I/O code in H5FD__mpio_write() to match that in H5FD__mpio_read(). Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Parallel rank0 deadlock fixes (#1183)jhendersonHDF2022-01-221-14/+39
| | | | | | | * Fix several places where rank 0 can skip past collective MPI operations on failure * Committing clang-format changes Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Only check HDF5_DRIVER in MPI I/O VFD if VFD isn't initialized (#1213)jhendersonHDF2021-11-251-13/+14
|
* Fix MPI Comm and Info leak in H5FD__mpio_delete (#1189)jhendersonHDF2021-11-121-0/+6
|
* Simplify function enter macros for performance benefits (#1024)David Young2021-11-081-29/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * Take a stab at using constructors to initialize instead of function-entry macros. This is a work in progress. It's good enough to run `many_dsets`. * Committing clang-format changes * Add the `many_dsets` benchmark and some scripts I used on jelly for setting up the build/test environment and for recording/flame-graphing profiles. * Committing clang-format changes * Change my Makefile and environment script to work both on jelly and on mayll (and probably on Summit). * Disable clang-format "fix." * Replace the `if (!H5_TERM_GLOBAL)` test in each FUNC_ENTER_ macro with `if (true)`. * Fix bad grammar in a comment. * Instead of labeling the H5*__init_package routines constructors, fold each into an initialization routine, H5*_init(), and call each of the H5*_init() routines. Call most of the H5*_init() routines from H5_init_library() in an explicit order that I found out earlier by instrumenting each __init_package routine and running the library tests. Roll H5FD*__init_package routines into H5FD*_init() routines. This change ends just-in-time initialization of package dependencies by package initializers. Don't track in per-package variables (H5_PKG_INIT_VAR) whether each package has been initialized. Instead, track in a single library variable whether the whole library is initialized or not. Drive the initialization of packages by H5_init_library() with a table of initializer routines. Also drive the termination of packages by H5_term_library() with a table. Perform initialization as needed from FUNC_ENTER_API_INIT(err). This basically restores the old behavior of that macro. Delete a bunch of #definitions in H5private.h that have fallen out of use with these changes. * Committing clang-format changes * Undo the bad auto-formatting that appears to have occurred in spite of my disabling it. Bracket some code in /* clang-format off */ /* clang-format on */ to prevent a recurrence. * Remove a diagnostic abort(). * Fix a logic error: print a comma between every package terminator run, and don't print an initial comma. * Complete the changes I started in H5_term_library() that undo the bad auto-formatting. Stop tracking whether package "tops" were initialized in per-package variables H5*_top_package_initialize_s. H5_term_library() takes care of that for them. Remove H5R_top_term_package() and H5R_term_package(), they don't do anything. * Committing clang-format changes * NFCI. Simplify macro text: replace `if (true) {` with `{`. * Fix formatting and suppress clang-format on a longer range. * Quiet some unused label, unused variable complaints that cropped up after I simplified the FUNC_ENTER_ macros for the sake of performance. * Committing clang-format changes * Delete some programs and scripts that don't belong in the pull request. * Use the right function-entry macro. * Use a sensible format and disable auto-formatting. * Stop calling do-nothing initializer H5FS_init(). Delete it. * Document what changes to make if the default VFD changes. * While I am here, change an `await_prior` flag on the terminator table to `true` to match the previous, non-table-driven code that was here. Found the oversight making the following changes: NFCI: insert an empty line and copy over slightly-edited comments from the previous version, where those comments still correctly explained how library termination operated. * NFCI: lower a staircase. * Replace every occurrence of FUNC_ENTER_NOAPI_INIT(...) with H5_PUSH_FUNC since that is all that that macro does any more. Quiet a bunch of new warnings by changing FUNC_ENTER_NOAPI(...) to FUNC_ENTER_NOAPI_NOERR and removing disused `done:` labels. * NFCI: add curly braces around a multiline statement. * Quiet a signed/unsigned comparison warning. * Add some documentation about library initialization and shutdown. * Make sure that the library is initialized, or else that initialization is already underway, before performing any VFD's initialization. * Committing clang-format changes * Committing clang-format changes * Reduce differences from `develop` branch. * Always initialize `tot_init`. * Committing clang-format changes * Fix typo: H5SL_init initializes skip lists, not VOL. * Remove H5_TERM_GLOBAL test in H5T_init. H5T_init was unusual in that it tested H5_TERM_GLOBAL and exited early if it was set. No other module initializers did that, and I cannot find any reason that should be necessary. Tests still pass when I remove it, so away it goes. * Use HD prefix. * Add function header comments. * Drop the intermediate variable, it's only used once. * Extract subroutine `H5FDperform_init(hid_t (*init)(void))` that initializes the library, if necessary, before calling its VFD-initializer argument. Use H5FDperform_init in the definition of the symbols H5FD_<vfd> (e.g., H5FD_SEC2), which may be evaluated before the library is initialized, like so: ``` ``` I implement H5FDperform_init in its own source file, H5FDperform.c, and exclude that file from trace processing because the `bin/trace` cannot deal with the function-pointer type. * Straggler from last: add new source file src/H5FDperform.c. * Committing clang-format changes * Add a missing file to the MANIFEST. * Switch to FUNC_ENTER_API_NOINIT in H5FDperform_init() and hbool_t in H5_term_library(). Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* VFD plugins (#602)jhendersonHDF2021-09-291-11/+52
| | | | | | | | | | | | | | | | | * Implement support for loading of Virtual File Drivers as plugins Fix plugin caching for VOL connector and VFD plugins Fix plugin iteration to skip paths that can't be opened * Enable dynamic loading of VFDs with HDF5_DRIVER environment variable * Temporarily disable error reporting during H5F_open double file open * Default to using HDstat in h5_get_file_size for unknown VFDs * Use macros for some environment variables that HDF5 interprets * Update "null" and "ctl testing" VFDs
* VFD ctl feature (#981)jhendersonHDF2021-09-231-117/+106
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * Added "ctl" callback to the VFD interface, and the associated H5FDctl() and H5FD_ctl() calls. Modified the MPIO VFD accordingly -- specifically: Added ctl() call with op-code support to expose rank, size, and communicator. Modified H5FD_mpi_get_rank(), H5FD_mpi_get_size(), and H5FD_mpi_get_comm() to use the new ctl() callback. In passing removed the const qualifier from the file parameter of these functions, as the file parameter of the ctl callback is not const. Deleted the old H5FD__mpio_mpi_rank(), H5FD__mpio_mpi_size(), and H5FD__mpio_communicator() calls from the MPIO VFD. Deleted H5FD_class_mpi_t from H5FDprivate.h, and modified the MPIO VFD accordingly. Note that all VFDs now use H5FD_class_t, with no special class for VFDs that that support MPI. Some minor touch ups to the Neil's selection I/O mods in passing. Tested serial and parallel, debug and production on charis and jelly. * Reserve a range of VFD "ctl" opcodes for library and experimental usage * Add "ctl" callbacks to passthrough VFDs * Add RELEASE.txt entry for "ctl" callback * Use H5FDopen with H5F_ACC_RDWR flag instead of H5F_ACC_TRUNC in vfd test * Remove handling of passthrough "ctl" flag from multi VFD * Move logic for testing H5FD_CTL__TEST_OPCODE into a testing VFD Revise description of "ctl" callback in RELEASE.txt Remove unused H5FD_CTL__NUM_OPCODES definition Fix some warnings in multi VFD Co-authored-by: mainzer <mainzer#hdfgroup.org>
* Fixes parallel issues from recent C99 changes (#809)Dana Robinson2021-06-301-28/+28
| | | | | | | * Fixes parallel issues from recent C99 changes * Adds MPE FUNC --> __func__ changes missed in earlier PRs * Even more missed FUNC --> __func__ macros
* Fix H5FD_mpio_Debug="rw" can report MPI_count instead of bytes (#699)Scot Breitenfeld2021-05-281-12/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | * fixed missed closing of a dataset * fixed missed closing of a dataset * fixed typo in error return * Committing clang-format changes * minor edits * code format * Committing clang-format changes * code format * minor edit * switched from using MPI_count, to actual bytes written for H5FD_mpio_debug rw debugging * Committing clang-format changes * changed size_i in printf to reflect the I/O. * Committing clang-format changes Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Brings the native implementation of H5Fdelete() from Bitbucket (#524)Dana Robinson2021-04-161-35/+91
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * Committing clang-format changes * Brings the native VFD H5Fdelete() implementation from Bitbucket Only brings the 'del' callbacks, not the 'open/close' scheme. * Formatter changes * Committing clang-format changes * Fixes direct VFD callback name * Removes UNUSED macro from family API call * Adds barrier and rank 0 check to MPI-I/O VFD delete * Revert "Adds barrier and rank 0 check to MPI-I/O VFD delete" This reverts commit 909765f759d9d96e84f4b8b1cc14f7d2b3ac8143. * Revert "Revert "Adds barrier and rank 0 check to MPI-I/O VFD delete"" This reverts commit 9b04bef1157853fc79fcb8fcc3e8ba1371091702. * Adds a second barrier after the delete in MPI-I/O VFD * Only delete files in the core VFD when the backing store flag is set * Fixes string issues in multi VFD Also, h5test.c cleanup code now uses H5Fdelete(). * Formatted source * Rework fapl checks for MPI-I/O VFD delete callback Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Clean up MPI-IO VFD tracing support (#520)Quincey Koziol2021-03-261-128/+196
| | | | | | | * Clean up tracing support * Committing clang-format changes Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Align with "parallel fence" changes (#479)Quincey Koziol2021-03-241-38/+45
| | | | | | | | | * Small code updates and simplifications * Committing clang-format changes * Fix 'make installcheck' for parallel builds Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Update license url (#332)Larry Knox2021-02-171-1/+1
| | | | | | * Modify temporary rpath for testing in java example scripts. * Update URL in source file Copyright headers for web copy of COPYING file - src and test directories.
* develop revert source to clang-format version 11 (#293)Allen Byrne2021-01-291-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * OESS-98 convert plugin option to FetchContent, add tests * Fixes for pkcfg files because of plugin option * OESS-98 fix tools test for plugins * Keep doxygen comments under 100 chars long - format hint * Whitespace * HDFFV-11144 - Reclassify CMake messages * HDFFV-11099/11100 added help text * Reworked switch statement to compare string instead * Fix typo * Update CDash mode * Correct name of threadsafe * Correct option name * Undo accidental commit * Note LLVM 10 to 11 format default changes * Update format plugin * Undo clang-format version 11 changes * One more correction
* H5FD__mpio_read: Fix MPI_Bcast datatype (#205)Evan Ramos2020-12-181-1/+5
| | | Signed-off-by: Evan Ramos <evan@hpccharm.com>
* Basic alignment with async branch (#115)Quincey Koziol2020-11-231-2/+2
| | | | | | | * Basic alignment with async branch - trivial changes to reduce clutter in overall diff. * Update minor error code to reflect change within library * Update the error output to match library
* Fixes minor issues in various virtual file driversDana Robinson2020-10-291-29/+0
| | | | | | | | | | | | - HDFS VFD files now ignored in the Autotools when the HDFS VFD is not being built. - All VFD init code uses idiomatic H5I_INVALID_HID instead of -1 or FAIL. - The HDFS VFD now includes H5FDdrvr_module.h in the right place. - The HDFS tests in test/hdfs.c no longer pass NULL to VERIFY macros, which raised warnings. - Minor tweaks to fix const, unused variables, etc. warnings in several VFDs.
* Clang-format of source filesAllen Byrne2020-09-301-406/+384
|
* Revert " Checkin of fix for CGNS bug"Quincey Koziol2020-08-271-3/+1
| | | | This reverts commit 94c34773ceae5b30c4afb227d0385ebf4ab6ce28.
* Clean up private / package / static namespace issues (function naming, whichQuincey Koziol2020-08-061-1/+1
| | | | | | header file, FUNC_ENTER / LEAVE, etc). Removed remaining personal email addresses from library source code (still needs cleaned from other directories). Misc. warning, style, and whitespace cleanup.
* Normalization with hdf5_1_10Dana Robinson2020-06-011-1/+1
|
* Trim trailing whitespaceQuincey Koziol2020-04-201-15/+15
|
* Another cleanup pass as suggested by the reviewers.Richard Warren2019-10-071-1/+0
|
* At the suggestion of the PR reviewers, moved the mpio_create_large_type to ↵Richard Warren2019-10-071-1/+1
| | | | H5mpi.c and renamed the function appropriately. Also moved some support functions to set and get the vvalue where we transistion to using derived datatypes.
* No functional changes, just removed some tab charactersRichard Warren2019-10-051-4/+4
|
* Make the initial bug fixes to allow >2GB writes with Independent IORichard Warren2019-10-051-10/+28
|
* Minor tweaks to new H5P MPI code based on code review feedback.Dana Robinson2019-08-271-14/+5
|
* Added H5Pset/get_mpi_params calls and unified them with the MPI-I/ODana Robinson2019-08-261-288/+91
| | | | VFD info in H5FDmpio.c.
* Some refactoring prior to implementing new H5P MPI functions.Dana Robinson2019-08-141-321/+316
| | | | | | * Macro cleanup and obvious warning fixes in parallel code. * Moved H5FD comm and info dup/free wrapper code to a new H5mpi.c file and separated it to deal with each MPI type separately.
* Added an H5MM_memcpy call that checks for buffer overlap.Dana Robinson2019-03-161-1/+1
|
* HDFFV-10625 -- Implemented a process-0 read and then broadcast for ↵Quincey Koziol2019-01-071-6/+37
| | | | collective read of full (HS_ALL), contiguous, atomic datasets by all the processes in the file communicator.
* Clean up style issues, make function callbacks static, correct debug statements,Quincey Koziol2019-01-051-475/+319
| | | | etc. No functionality changes at all.
* Merge branch 'develop' into vfd_init_cleanupDana Robinson2019-01-011-0/+1
|\
| * Clear driver properties before returning them to application.Quincey Koziol2018-12-191-0/+1
| | | | | | | | | | Move write_tracking and page_size properties to be inside the core VFD's driver properties.
* | Cleaned up and normalized VFD init calls.Dana Robinson2018-12-181-16/+18
|/
* Fixed HDFFV-10404Binh-Minh Ribler2018-07-131-2/+2
| | | | | | | | | Description: Applied the typo fixes from user's report. The previous pull request couldn't be merged because it was too old, and it was too complicated for me to resolve conflicts. Platform tested: Linux/64 (jelly) - very minor
* Add an assert to verify that we don't write when between MPI_Barrier() callsQuincey Koziol2018-03-271-0/+3
| | | | during a flush.
* Merge branch 'develop' of https://bitbucket.hdfgroup.org/scm/hdffv/hdf5 into ↵Quincey Koziol2018-03-181-23/+74
|\ | | | | | | | | | | merge_func_enter_vol Plus initial steps toward merging API context push into FUNC_ENTER_API* macros
| * Removed commented out code from H5FDmpio.c, H5FDprivate.h and H5Fint.cmainzer2018-03-121-88/+0
| | | | | | | | Tested parallel (debug and production) and serial (production)
| * Optimized version of avoid truncate patch.M. Scot Breitenfeld2017-12-181-0/+198
| |
* | Add API context interface and use it throughout the library.Quincey Koziol2018-03-151-46/+22
|/
* Fixed HDFFV-10214:Dana Robinson2017-05-191-4/+5
| | | | | | | | | | | | * Updated comments in the library to accurately reflect the semantics of the H5FD_FEAT_POSIX_COMPAT_HANDLE flag. * Removed the check for the POSIX compatible handle feature flag when determining if a VFD supports SWMR. Partial work for HDFFV-10197: * Added a new H5FD_FEAT_DEFAULT_VFD_COMPATIBLE feature flag that is set when a VFD creates output that is compatible with the default VFD. Will be used in the testing but might also be generally useful to users.
* Merge pull request #426 in HDFFV/hdf5 from ~LRKNOX/hdf5_lrk:hdf5_1_10 to ↵Larry Knox2017-04-251-6/+4
| | | | | | | | | | hdf5_1_10 * commit '54957d37f5aa73912763dbb6e308555e863c43f4': Commit copyright header change for src/H5PLpkg.c which was added after running script to make changes. Add new files in release_docs to MANIFEST. Cimmit changes to Makefile.in(s) and H5PL.c that resulted from running autogen.sh. Merge pull request #407 in HDFFV/hdf5 from ~LRKNOX/hdf5_lrk:hdf5_1_10_1 to hdf5_1_10_1 Change copyright headers to replace url referring to file to be removed and replace it with new url for COPYING file.
* Checkin of fix for CGNS bugmainzer2017-04-061-1/+36
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | (https://jira.hdfgroup.org/browse/HDFFV-10055). Briefly, in H5C_collective_write() in H5Cmpio.c, the metadata cache attempts to perform a collective write of metadata cache entries. This worked fine as long as all processes had at least one entry to write. However, when the process has no entries, the function tries to participate in the collective write by calling MPI_File_set_view(), MPI_File_write_all() and then MPI_File_set_view() again, to match the calls in H5FD_mpio_write(). After pull request 183, the CGNS test benchmark_hdf5 started failing. On investigation, I determined that the failure occurred in the first call to MPI_File_set_view() in the "no data to write" path through H5C_collective_write(). Note that pull request 183 did not create the problem, it only exposed it. The bug can be observed after pull request 182 if one executes the CGNS progam src/ptests/benchmark_hdf5 with 90 processes. The problem appears to have been that the calls to MPI_File_set_view() in H5C_collective_write() and H5FD_mpio_write() were using different values for the info parameter. I patched the problem by adding a MPI specific VFD call allowing me to get the MPI_Info used in H5FD_mpio_write() for use in MPI_File_set_view() calls in H5C_collective_write(). Tested serial & parallel, debug & production on Jelly.
generated by cgit v0.12 at 2024-07-09 09:27:17 (GMT)