Problem appears to have been caused by file system contention.
In the chunked dataset case, reshaping the chunks so that only one
process would touch each chunk and setting the alignment equal to the
default Lustre block size more or less dealt with the problem.
For contiguous datasets, the problem was a bit more difficult, as
re-working the test to avoid contention would have been very time
consuming.
Instead, I added code to time one execution of each type of shape
same test, and skip additional tests of that type if the duration
of the test exceeded some threshold
In all cases, I set up code to turn off the above fixes if express
test is 0.
Tested on Abe and commit tested. On the commit test, the configure
test failed -- probably because I was h5committest from heiwa due
to some ssh wierdness. In any case a manual reconfigure run on
jam seemed to work fine.
Also, in h5committest, I ran into some data conversion warnings.
I didn't worry about them as the only code I changed was in testpar.
Add missing files from the previous checkin.
(Add extra test cases for Bug1975 h5diff - support recursive comparison on group when specified as an object)
Add extra test cases for bug1975 h5diff - support recursive comparison on group when specified as an object.
Description:
Additional tests for combination of group recursive and --follow-symlinks
with multi-linked external links with several files (same name/strucure).
Tested:
jam, amani, heiwa
Add --exclude-path option
Description:
Specified path to an object will be excluded from comparing the two files or two groups. If group is specified all the member objects will be excluded.
Related to "1890: h5diff excluding object for file comparison via command line"
Tested:
jam, amani and heiwa
Correct Bz#1968 (Problem when opening new format file with
H5F_LIBVER_EARLIEST) by initializing symbol table leaf information in a
more universal way. Also cleaned up other superblock initialization a
little bit while I'm here.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
FreeBSD/64 6.3 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (jam) w/PGI compilers, w/default API=1.8.x,
w/C++ & FORTRAN, w/threadsafe, in debug mode
Linux/64-amd64 2.6 (amani) w/Intel compilers, w/default API=1.6.x,
w/C++ & FORTRAN, in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, w/threadsafe, in production mode
Linux/PPC 2.6 (heiwa) w/C++ & FORTRAN, w/threadsafe, in debug mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Linux/64-amd64 2.6 (abe) w/parallel, w/FORTRAN, in debug mode
Mac OS X/32 10.6.4 (amazon) in debug mode
Mac OS X/32 10.6.4 (amazon) w/C++ & FORTRAN, w/threadsafe,
in production mode
Mac OS X/32 10.6.4 (amazon) w/parallel, in debug mode
Fix for Bug1975 h5diff - support recursive comparison on group when specified as an object
Description:
Compare member objects and groups recursively when two files or groups are specified to be compared. Support parallel diff and handling symbolic links accordingly.
Tested:
jam, amani, heiwa
Clean up the internal usage H5I_dec_ref() to eliminate the sequence of
flags that has been creeping in.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
FreeBSD/64 6.3 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (jam) w/PGI compilers, w/default API=1.8.x,
w/C++ & FORTRAN, w/threadsafe, in debug mode
Linux/64-amd64 2.6 (amani) w/Intel compilers, w/default API=1.6.x,
w/C++ & FORTRAN, in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, w/threadsafe, in production mode
Linux/PPC 2.6 (heiwa) w/C++ & FORTRAN, w/threadsafe, in debug mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Linux/64-amd64 2.6 (abe) w/parallel, w/FORTRAN, in debug mode
Mac OS X/32 10.6.4 (amazon) in debug mode
Mac OS X/32 10.6.4 (amazon) w/C++ & FORTRAN, w/threadsafe,
in production mode
Mac OS X/32 10.6.4 (amazon) w/parallel, in debug mode
Bring back various modifications to the v2 B-trees that were introduced
in the revise_chunks branch but aren't dependent on any file format or API
changes.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
FreeBSD/64 6.3 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (jam) w/PGI compilers, w/default API=1.8.x,
w/C++ & FORTRAN, w/threadsafe, in debug mode
Linux/64-amd64 2.6 (amani) w/Intel compilers, w/default API=1.6.x,
w/C++ & FORTRAN, in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, w/threadsafe, in production mode
Linux/PPC 2.6 (heiwa) w/C++ & FORTRAN, w/threadsafe, in debug mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Linux/64-amd64 2.6 (abe) w/parallel, w/FORTRAN, in debug mode
Mac OS X/32 10.6.4 (amazon) in debug mode
Mac OS X/32 10.6.4 (amazon) w/C++ & FORTRAN, w/threadsafe,
in production mode
Mac OS X/32 10.6.4 (amazon) w/parallel, in debug mode
Description:
After a C++ API object is closed, its id should be reset regardless
of the reference count of the underlying HDF5 object. User reported
the bug for Group, but it's applicable to all other objects. Fixed.
Platforms tested:
Linux/32 2.6 (jam)
FreeBSD/64 6.3 (liberty)
SunOS 5.10 (linew)
Additional fix for Bug 1821 - h5repack outputs compression information where it is not supposed to
Description:
Remove unnecessary lopping code, which only lowers performance.
Tested:
jam, amani, heiwa
Fix for Bug1896 h5repack - changing layout to COMPACT does not work
Description:
Make h5repack be able to convert a layout to COMPACT for small size dataset as default. Also add verifying layout changes in our test script.
Tested:
jam, amani, heiwa
Description:
The fractal heap header structure keeps a pointer to the file associated with
it. However, it is possible for that file pointer to be closed while the
header is still in cache (through the shared file pointer). Previously, the
header's file pointer was not updated and subsequently pointed to an invalid
file structure. Modified fractal heap code to update the file pointer every
time the header is accessed.
Tested: jam, linew, amani (h5committest)
Correct an error I convinced Mike to introduce with the phread_once()
code, :-( along with another one that wasn't my fault. Also, clean up
warnings and restructure code in the thread-safe code a bit.
Tested on:
Mac OS X/32 10.6.4 (amazon) w/production + thread-safe
(not a configuration that h5committest tests)
Add windows threads support to HDF5.
Description:
Added calls to the windows threads library to the H5TS layer, and wrapped
most calls to either pthreads or windows threads library with portable
H5TS-style defines. Modified tests to use portable function definitions
as well.
This can be configured via CMake with the HDF5_ENABLE_THREADSAFE
option, and should work on windows vista and later operating systems.
Tested:
h5committest, plus threadsafe with pthreads on jam and amani,
and tested on a Windows Vista VM with threadsafe using windows threads.
Description:
The development (--disable-production) mode ran a long time, especially on the
testhdf5 (select) that exceeded the wall clock 10 minutes limits and was
terminated. It turned out the development had DEBUG_CFLAG set to "-g -xildoff".\
The -xildoff was added back when h5detect had errors when -g only was used. But
that bug does not exist any more since when only -g is used to build, h5detect
did not fail and all tests passed. Therefore -xildoff is removed.
Furthermore, when "-g -O" is used to build the library, debuggers such as
dbx can still use the executable, though some debugging ability like displaying
local variables is not available in optimized code. But overall execution time
shorten.
Therefore, the C, Fortran aand CXX DEBUG flags are all set to "-g -O" by
default.
Tested: in Linew using the default development mode because the changes affect
only the Sun platform.
Description:
testhdf5/select runs very slow when in --disable-production mode using
pgcc compiler. This was caused by the -Mchkfpstk compiler switch added
to the DEBUG_FLAGS in config/pgi-flags file.
Solution:
Removed -Mchkfpstk from both pgi-flags and pgi-fflags.
See Bug 2026 for dettails.
Tested: (no h5committest test since they don't use PGI compilers.)
Jam (linux32): using pgcc/pgf90/pgCC using default and --enable-production.
Amani (linux64):
using pgcc/pgf90/pgCC (32 bits mode by "-tp k8-32"), using default and --enable-production.
Add some extra range checking for invalid chunk sizes.
Tweak default chunk sizes to be zero, so that it's more likely that
application developers will be forced to set them correctly.
Add a few tests to verify these checks.
Tested on:
Mac OS X/32 10.6.4 (amazon) w/debug & production
(too minor to require h5committest)
Corrected use/name of source folder aliases.
Duplicated FindMPI.cmake so that non-c++ compiler is found first (recommemded commands did not work).
Tested: local linux with mpich
Bring recent tweaks to extensible and fixed array file formats from
revise_chunks branch back to trunk.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
FreeBSD/64 6.3 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (jam) w/PGI compilers, w/default API=1.8.x,
w/C++ & FORTRAN, w/threadsafe, in debug mode
Linux/64-amd64 2.6 (amani) w/Intel compilers, w/default API=1.6.x,
w/C++ & FORTRAN, in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, w/threadsafe, in production mode
Linux/PPC 2.6 (heiwa) w/C++ & FORTRAN, w/threadsafe, in debug mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Linux/64-amd64 2.6 (abe) w/parallel, w/FORTRAN, in debug mode
Mac OS X/32 10.6.4 (amazon) in debug mode
Mac OS X/32 10.6.4 (amazon) w/C++ & FORTRAN, w/threadsafe,
in production mode
Mac OS X/32 10.6.4 (amazon) w/parallel, in debug mode
Clean up a few compiler warnings and style issues in/near recent changes.
Tested on:
Mac OS X/32 10.6.4 (amazon) w/debug & production
(too minor to require h5committest)
Bump optimization down for Intel 10.* compilers, due to apparent bad cod
e generation.
Tested on:
Linux/64 2.6 (cobalt) w/debug & production & Intel 11.1 compilers
(h5committest not appropriate)
Bring revisions from Coverity branch back to trunk:
r19191:
Fix coverity items 104 and 105. Added calls to H5HF_sect_indirect_free to
H5HF_sect_indirect_reduce and H5HF_sect_indirect_reduce_row if there is an errorbefore "peer_sect" is linked into the main free space structure via its direct
sections. Also delayed call to H5HF_sect_indirect_first to prevent peer_sect
from being left in an inconsistent state.
r19268:
Added fix to disallow extendible compact dataset.
This was the same check as in H5D_contig_construct() in H5Dcontig.c.
Added test to verify the creation of extendible dataset with various layouts.
Tested on:
Mac OS X/32 10.6.4 (amazon) w/debug & production
(h5committested on branch)
