Since each API context is local to a thread, use the stack to
store the context instead of allocating & releasing it each time.
This improves performance (slightly), reduces alloc/free calls,
and eliminates the H5FL package from the push & pop operations,
which helps simplify threadsafe operation.
One effect of this change is that the H5VLstart_lib_state /
H5VLfinish_lib_state API routines for pass through connector
authors now require a parameter that can be used to store
the library's context. It was probably a mistake to assume
that these two routines would not do this previously, so this
is essentially a bug fix for them.
Some other minor things:
* Added API context push+pop operations to cache tests
(I'm not actually certain why this was working before) and
a few other places
* Cleaned up a bunch of warnings in test code (calloc args, mainly)
* Made header file inclusions more standard in some source files
This is where most people will expect to find license information. The
COPYING_LBNL_HDF5 file has also been renamed to LICENSE_LBNL_HDF5.
The licenses are unchanged.
Added setup and cleanup callback parameters to TestInit() to perform
setup and cleanup tasks once for whole test program
Removed TestCleanup() function since its functionality is covered by
PerformTests()
Added check of the HDF5_NOCLEANUP environment variable in
GetTestCleanup()
Added new testframe.h header to document testing framework functions and
split them away from h5test.h and from test programs that don't
integrate with the testframe.c testing framework
Added new test setup callback to testframe.c testing framework
Added parameters to AddTest() to specify size of test parameters so they
can be copied for later use
Enabled HDF5 error stacks in testframe.c framework by default and added
some error stack suppressions to some testhdf5 tests
Added new maxthreads option to testframe.c framework to allow specifying
the maximum number of threads a multi-threaded test can use
Moved TestExpress functionality out of testframe.c and into more general
h5test.c for wider use by tests through getter and setter
Updated some tests to not mix and match functionality between h5test.c/h
and testframe.c/h
Moved some functionality from testphdf5.h into testpar.h for parallel
tests that aren't part of testphdf5
Added new parallel test library that contains common shared
functionality for parallel tests (similar to h5test library)
Fixes a bug where the vector I/O sizes weren't being extended when
one of the entries in the array is 0. This caused an over-read of
the I/O sizes buffer and on some machines would cause a memory
allocation failure due to the calculated I/O vector size being too
large.
The t_pmulti_dset and t_select_io_dset tests rely on the behavior
of the previous private rand_r-like implementation to get the
correct sequence of random numbers to pass. This has been restored
using a fully private rand_r-like implementation that doesn't
rely on rand_r and will work on Windows and other platforms
where rand_r doesn't exist.
* Clean up random number generator code
Depending on the platform, we use a mix of random, rand, and rand_r
to generate pseudo-random numbers, along with a messy set of ifdefs
in H5private.h. We are not a cryptographic library, only use random
numbers in our test code, and have no need for anything more than the
C standard's (s)rand(). There's no point dithering about using rand()
vs random() when we're also doing bad things like using mod to
restrict the range, which introduces bias.
Also removes CMake/configure checks for rand_r and random
* Remove random/rand_r checks from build system
* Fix missed HDrandom after GitHub merge
Changes Autotools testing to use HDF5_TEST_DRIVER environment
variable to avoid running tests that don't work well with several
VFDs
Restores old h5_get_vfd_fapl() testing function to setup a FAPL
with a particular VFD
Adds a macro for the default VFD name
* Fix issue with Subfiling VFD and multiple opens of same file
* Update H5_subfile_fid_to_context to return error value instead of ID
* Add helper routine to initialize open file mapping
The buffers passed to stat-like calls are only partially filled in by
the call, leaving ununitialized memory areas when the stat buffers are
created on the stack.
This change memsets the buffers to 0 before the stat calls, quieting
the -fsanitze=memory complaints.
off_t is a 32-bit signed value on Windows, so we should use HDoff_t
(which is __int64 on Windows) internally instead.
Also defines HDftell on Windows to be _ftelli64().
Vector I/O requests are now processed within a single
set of I/O call batches, rather than each I/O vector
entry (tuple constructed from the types, addrs, sizes
and bufs arrays) being processed individually. This allows I/O to be
more efficiently parallelized among the I/O concentrator processes
during large I/O requests.
* Fixed some calculations and add test cases for issues spotted from review
* Removed a variable that was compensating for previous miscalculations
* Changes for ECP-344: Implement selection vector I/O with collective chunk filling.
Also fix a bug in H5FD__mpio_write_vector() to account for fixed size optimization
when computing max address.
* Fixes based on PR review comments:
For H5Dchunk.c: fix H5MM_xfree()
For H5FDmpio.c:
1) Revert the fix to H5FD__mpio_write_vector()
2) Apply the patch from Neil on the proper length of s_sizes reported by H5FD__mpio_vector_build_types()
* Put back the logic of dividing up the work among all the mpi ranks similar to the
original H5D__chunk_collective_fill() routine.
* Add a test to verify the fix for the illegal reference problem in H5FD__mpio_write_vector().
* Fix issue with HDF5_VOL_ALLOW_EXTERNAL CMake variable
* Add initial API test workflow
* Initialize parallel testing with MPI_THREAD_MULTIPLE when testing API
* Add CMake variable to allow specifying a VOL connector's package name
* Remove call to MPI_Init in serial API tests
While previously necessary, it now interferes with VOL connectors that
may need to be initialized with MPI_THREAD_MULTIPLE
* HDF5 API test updates
Removed test duplication from bringing API tests
back into the library from external VOL tests
repo
Synced changes between API tests and library's
tests
Updated API tests CMake code to directly use and
install testhdf5, testphdf5, etc. instead of
creating duplicate binaries
Added new h5_using_native_vol() test function to
determine whether the VOL connector being used
is (or the VOL connector stack being used resolves
to) the native VOL connector
* Remove duplicate variable
Allow H5Pset_evict_on_close to be called regardless of whether a parallel build of HDF5 is being used
Fail during file opens if H5Pset_evict_on_close has been set to true on the given File Access Property List and the size of the MPI communicator being used is greater than 1
This function allows the user to determine if the library performed selection I/O, vector I/O, or scalar (legacy) I/O during the last HDF5 operation performed with the provided DXPL. Expanded existing tests to check this functionality.
Broadcast number of datasets to create in multi-dataset I/O cases
so that interference with random number generation doesn't cause
mismatches between ranks
Set fill time for datasets to "never" by default and adjust on a
per-test basis to avoid writing fill values to chunks when it's
unnecessary
Reduce number of loops run in some tests when performing multi-dataset
I/O
Fix an issue in the "fill time never" test where data verification
could fill if file space reuse causes application buffers to be
filled with chosen fill value when reading from datasets with
uninitialized storage
Skip multi-chunk I/O test configurations for multi-dataset I/O
configurations when the TestExpress level is > 1 since those
tests can be more stressful on the file system
Disable use of persistent file free space management for now
since it occasionally runs into an infinite loop in the library's
free space management code
* Fix printf warnings in t_mpi
The type of MPI_Offset varies with implementation. In MPICH, it's long,
which raises warnings when we attempt to use long long format
specifiers. Casting to long long fixes the warnings.
jhendersonHDF
Allen Byrne
Quincey Koziol
Dana Robinson
H. Joe Lee
Glenn Song
Sean McBride
vchoi-hdfgroup
Scot Breitenfeld
Neil Fortner