could block at the barrier, wake and exit the barrier, re-acquire the barrier
lock and increase `nentered` before the other blocked threads woke and checked
`nentered % count == 0`. Then the other blocked threads would check `nentered
% count == 0` and, finding it false, go back to sleep in the barrier. This new
implementation waits for a looser condition to obtain so that threads don't go
back to sleep in the barrier.
* commit 'cf58730177d60fd7311582a29539d87f10a7d88e':
Update Windows platforms
HDFFV-11036 add release note
HDFFV-11036 add file compare test process
Correct usage of add_compile_definitions
* commit '380fe7cfdfd98461cb1bff54bad4de0ee9904ad6':
Fix always true issue because member is not dynamically allocated
CMake cleanup
HDFFV-11032 fix JNI call
* commit '158ba17be0ad40e3e47a951c1340710608347772':
Replace the old H5F_DECODE_LENGTH() implementation with one that initializes the variable it's loading on every path.
* commit '50aac126d7c4082c2aeb0f3a68cad26bd1cd9783':
Start to document the function/parameter/variable attribute macros.
Filter all of the new _USED / _UNUSED type modifiers.
H5_HAVE_PARALLEL, _DEBUG_API, etc. Add attributes to some variables and parameters that are unused under some configurations.
Merged per discussion with Elena.
* commit 'b4697f969295245840350e17d75d92a0fdd7e7a9':
Use HD prefix.
Delete unhelpful comment per Jordan's question.
Make sure that H5TS_thread_id() is available as either a function or a macro in all configurations.
Provide local copies of err(3)- and errx(3)-alike functions for Visual Studio compatibility.
Provide C99/POSIX.1 format strings PRI[doux]{8,16,32,64,MAX,PTR} on systems that are missing <inttypes.h>.
Merged per discussion with Elena.
* commit '3f903a441ad84001ea66589728bd8b036b6fdfca':
Take out the temporary performance tests.
Make calls through a function pointer. Use the same number of arguments, always.
Increase iterations, provide a baseline for no-op, simplify the overhead case a bit.
Temporarily add some code that measures the time to run the simplest possible H5T__copy_all()-like routine 10 million times and then measures the version with FUNC_ENTER_STATIC/_LEAVE_NOAPI and a HGOTO_ERROR() statement.
H5T_copy() constification plus Quincey's contributions.
* commit '1b937c69f4b8187401eeb8eef195bbcc2ea58312':
HDFFV-11014, fix the h5repack issue that misses a few attributes during the repacking. The flag that checks the object reference attribute is not updated properly. The fix is trivial. Just need to move the flag update line into the inner loop. Tested at Jelly. Also update the release.txt.
all configurations.
Previously it was neither declared nor defined in --disable-threadsafety
builds. The compiler's warning got lost in the noise---I first saw the issue
because my -Werror branch stopped compiling cold---and the tests still linked
and ran.
