| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
See Kent's documentation "Designed to Fail Tests and Issues".
(a) Fix for issue #2:
--Print out meaningful message about max_lag when there is
checksum error in loading an entry via H5C__load_entry().
--H5C.c: H5C_protect()
(b) Fix for issue #4:
--Allocate more space when the copy of the index read from the metadata file is
bigger than the existing size
--H5Fvfd_swmr.c: H5F_vfd_swmr_reader_end_of_tick()
(B) When putting the old index into the delayed free list, use the old
writer_index_offset instead of the current writer_index_offset
--H5Fvfd_swmr.c: vfd_swmr_enlarge_shadow_index()
(C) When there is error form calling H5F_vfd_swmr_process_eot_queue() in
VFD_SWMR_ENTER(err) and VFD_SWMR_LEAVE(err), should report
FAIL instead of "err"
--H5private.h: VFD_SWMR_ENTER and VFD_SWMR_LEAVE macros
(D) Add tests to verify issue #2 and issue #4 are fixed.
|
|
|
|
|
|
|
|
| |
See Kent's documentation "Designed to Fail Tests and Issues".
(A) Fix for issue #1: HDassert the < and = cases between old and new
entry length. John will take care of the > case.
(B) Fix for issue #3: set the cache copy of page_size again if different
from f->shared->fs_page_size.
|
| |
|
|
|
|
|
| |
1. added an option to enable the legacy SWMR in vfd_swmr_bigset_writer.c.
2. adjusted the options for big set test to make sure it passes the exhaustive test in testvfdswmr.sh.in.
|
| |
|
| |
|
|
|
|
| |
new command-line options.
|
|\
| |
| |
| | |
raylu_bigset_test
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
(1) test/vfd_swmr_attrdset_writer.c
Free memory for variable length string + cleanup
(2) test/vfd_swmr_bigset_writer.c
Free memory for s.dataset and s.sources
(3) test/vfd_swmr_vlstr_reader.c
Free memory for content[]
Also fix a bug in test/testvfdswmr.sh.in so that os_groups_seg test will only be executed when specified.
|
| |\
| | |
| | |
| | | |
https://github.com/vchoi-hdfgroup/hdf5 into noflush_feature_vfd_swmr
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
test/vfd_swmr_group_writer with one million old style groups.
(A) src/vfd_swmr_enlarge_shadow_index():
--When enlarging the metadata file index entries, HDmalloc() is used to allocate space for the index entries,
but they are later freed at file close via H5MM_xfree() in H5F__dest().
This is corrected to use H5MM_calloc() instead to be consistent. Note that entries are initially allocated
also via H5MM_calloc() in H5F__vfd_swmr_create_index().
--If the pointer to the old metadata file index entries exists (shared->old_mdf_idx), need to free
the old entries via H5MM_xfree().
(B) test/testvfdswmr.sh.in:
--The test is added to verify that the bug is resolved.
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
src/H5Fvfd_swmr.c
--enable or disable raw data flush depending on configuration
(2)
vfd_swmr_dsetchks_writer.c
vfd_swmr_dsetops_writer.c
--modifications to test/verify when flushing of raw data is enabled/disabled
--refactor coding
(3)
testvfdswmr.sh.in
--add testing with/without flush of raw data to tests: dsetchks and dsetops
(4)
vfd_swmr_attrdset_writer.c
--pass true for flushing raw data to the common routine init_vfd_swmr_config()
--refactor coding so that it has similar format as the other two tests: dsetops and dsetchks
(5)
vfd_swmr_addrem_writer.c
vfd_swmr_bigset_writer.c
vfd_swmr_generator.c
vfd_swmr_group_writer.c
vfd_swmr_reader.c
vfd_swmr_remove_reader.c
vfd_swmr_remove_writer.c
vfd_swmr_sparse_reader.c
vfd_swmr_sparse_writer.c
vfd_swmr_vlstr_reader.c
vfd_swmr_vlstr_writer.c
vfd_swmr_writer.c
vfd_swmr_zoo_writer.c
--pass true for flushing raw data to the common routine init_vfd_swmr_config()
|
| | |
|
|\ \
| |/
| |
| | |
raylu_bigset_test
|
| | |
|
| |
| |
| |
| |
| |
| | |
(1) Change choose_dataset() to print out related info based on verbosity.
(2) Modify all routines in legacy and VFD SWMR tests that call choose_dataset().
(3) Modify testvfdswmr.sh.in to run tests based on HDF5TestExpress setting so as to reduce running time.
|
|/
|
|
|
|
| |
1. changed the signal handling to named pipes for communication between the writer and reader;
2. the writer writes as many chunks within a tick then notify the reader to verify the data;
3. some refactoring work.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Brings CMake updates from develop
* Brings reduction in pedantic casts from develop
* Purges UFAIL from the library (#637)
* Committing clang-format changes
* Purges UFAIL from the library
* H5HL_insert change requested in PR
Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Removes gratuitous (double)x.yF casts (#632)
* Committing clang-format changes
* Removes gratuitous (double)x.yF casts
* Committing clang-format changes
Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Committing clang-format changes
* Cleans up a const warning left over from previous constification (#633)
* Committing clang-format changes
* Adds consts to a few global variables
* Cleans up a const warning left over from previous constification
Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Formatted source
* Bring over some VOL registration changes
* VOL cleanup
* H5VL_request_status_t substitutions
* H5F.c cleanup
* Minor API tweaks from develop
* Moves H5G package init/teardown to H5Gint.c
* H5G cleanup
* H5M cleanup
* H5SM cleanup
* H5T cleanup
* H5R cleanup
* H5Lpublic.h cleanup
* H5L cleanup
* H5O cleanup
* H5A, H5CS, and H5AC cleanup
* Moved H5A init/teardown code to H5Aint.c
* Moves H5D I/O functions to H5D.c
* H5D cleanup
* Misc minor cleanup
* H5P close cleanup
* H5Tpublic.h cleanup
* Fixes err_compat test
* H5PLpublic.h cleanup
* Updates H5Ppublic.h
* H5Fpublic.h updates
* H5A.c cleanup
* Brings over H5Aexists and related changes
* Brings CMake shell testing changes from develop
* Close callback changes
* H5R and H5Tcommit normalization
* err_compat test works now
* H5O tweaks
* Updates VOL registration code
* Brings over H5VL_create_object
* H5Tconv.c reformatting
* H5T.c tweaks
* Brings datatype and reference updates from develop
* Brings VOL plugin loading changes from develop
* Brings event sets from develop
* Brings async functions over
* Tools changes
* Brings over many tools changes from develop
* Brings VOL flags from develop
* Fixes h5dump double/float tests
* Updates h5repack tests
* Brings h5diff test changes from develop
* Last h5dump changes
* Brings test changes from develop
* Committing clang-format changes
* Tidied h5_testing()
* Brings chunk iteration code + misc from develop
* Updates vds test
* Enables external link parallel test
* Brings updated property lists from develop
* H5G changes from develop
* H5MF cleanup
* Brings vfd_swmr test back into CMake
* Updates threadsafe test
* Updates plist test
* Brings recent changes from develop
* Require a C++11 compiler to build the C++ wrappers (#693)
* Committing clang-format changes
* Requires a C++11 compiler to build the C++ wrappers
Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Fix H5FD_mpio_Debug="rw" can report MPI_count instead of bytes (#699)
* Removes pre-C99 build and header cruft (#700)
* Assumes ANSI C headers exist
* Assumes stdbool.h, stdint.h, and inttypes.h are present
* Assumes the C++ compiler can handle stdint.h
* Removes all work-arounds for missing functionality, especially stdbool.h
Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
* Updates release documents
* Brings parallel changes from develop
* Adds a VFD/parallel check program for the VFD SWMR shell test
* Updates MANIFEST
* Disable building VFD SWMR acceptance tests on CMake
* Removes err.h from genall5.c
Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Dataset types:
--dataset with compact layout
--dataset with contiguous layout
--datasets with chunked layout: single/implicit/fixed array/extensible array/btree2 indexes
Types of writes:
--Sequential writes
--Random writes
--Regular hyperslab writes
--Raw data modifications
|
| | |
|
|\ \
| |/ |
|
| | |
|
| |
| |
| |
| | |
Also switches bin/bash shebang to use env instead
|
|/
|
|
|
| |
move_group, insert_links, delete_links, link storage transition and old-styled group
creation.
|
|
|
|
|
| |
2. Change the test names and remove unnecessary debugging information
Tested at jelly.
|
| |
|
| |
|
|\ |
|
| |\
| | |
| | | |
Feature/vfd swmr
|
| | |
| | |
| | |
| | | |
It is to test attribute handling for different dataset types.
|
| | | |
|
| | | |
|
| |/
|/|
| |
| |
| |
| | |
tests are fine. Need to clean up code and add comments/descriptions.
2. Add group attribute tests to testvfdswmr.sh.in. Only use typical cases for the standard and the quick runs for saving time.
|
|\ \
| |/
|/| |
Changed the signal handling to named pipes in the zoo test
|
| | |
|
| |
| |
| |
| |
| |
| | |
vfd_swmr_reader_did_increase_tick_to) that overrode the functions of the same names in the SWMR driver.
Also added a command-line option to disable named pipes so that the writer and reader can run seperately.
|
| |
| |
| |
| | |
and reader through the named pipes; using the error report consistent with other tests.
|
| |\
| | |
| | |
| | | |
raylu_zoo_test
|
| |\ \ |
|
| | | |
| | | |
| | | |
| | | | |
writer and reader, mainly in the zoo test as a starting point.
|
|\ \ \ \
| | |_|/
| |/| |
| | | | |
raylu_group_test2
|
| | | | |
|
| | | | |
|
| | | |
| | | |
| | | |
| | | | |
level of HDF5TestExpress, to speed up the test or do more thorough test.
|
|\ \ \ \
| |/ / /
| | | |
| | | | |
raylu_group_test2
|
| | |/
| |/| |
|
|/ /
| |
| |
| |
| | |
1. Reader waits for a number of ticks (3 by default) before verifying the creation of groups and attributes;
2. Error handling is consistent with traditional setup of the test suite.
|
|/
|
|
| |
Default, Exhaustive, and quick runs.
|
| |
|