Bring r19297 from trunk to 1.8 branch:
Whack a few more memory errors exposed by valgrind.
Make the debugging dump output a little bit prettier.
Tested on:
FreeBSD/32 6.3 (duty) w/debug
(h5committested on trunk)
Avoid getting object information for soft/external links if we aren't
going to traverse across the link itself.
Tested on:
Linux/64 2.6 (chicago) w/debug & production
(h5committest coming up)
Correct another error in metadata accumulator dirty region calculations
(this time with a corner case when freeing data in the file).
Tested on:
Linux/32 2.6 (talwit) w/debug
(following up with h5committest)
Description: aix -q32 default to a memory limit that some HDF5 tests would
fail in malloc.
Solution:
Temporary hardset
RUNSERIAL=${RUNSERIAL="env LDR_CNTRL=MAXDATA=0x20000000@DSA"}
to bypass this problem until the test programs are fixed. Though -q64 binary
has a higher default memory limit and does not need this patch, it is applied
the same for the sake of simplicity since it is a temporary fix.
Tested:
in BP, both -q32 and -q64 modes. No h5committest since this applies to the
AIX platform only.
Bring r19278 from trunk to 1.8 branch:
More fixes to memory allocation, etc. exposed by valgrind.
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
H5T_ORDER_MIXED specifically for compound types and its derived types. They report mixed
orders if they have members of different orders.
Tested on heiwa, jam, and amani.
Bring r19276 from trunk to 1.8 branch:
Correct calculation of end of dirty information in metadata accumulator
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
(h5committested on trunk)
Bring r19272 & 19273 from trunk to 1.8 branch:
Close out various resource leaks and usages of uninitialized memory that
were flagged by valgrind. [There's still some more valgrind warnings, but
it's better now... :-/ ]
Also clean up warnings and code formatting.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
FreeBSD/64 6.3 (liberty) w/C++ & FORTRAN, in debug mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production 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
Corrected use of "SKIP" message in cmake.
Corrections to cmake configuration files (for cmake external projects).
Community suggested changes.
Tested: local linux
Bring r19252 & 19253 from trunk to 1.8 branch:
Bring Coverity changes from branch to trunk:
r19161:
Fixed the part for matching the subset info with dataset
r19189:
BZ1646: h5dump does not check number of dimensions for subsetting parameters against the dataset
Changed subset_t structure from holding hsize_t pointers to holding new subset_d pointers, which hold the original hsize_t pointer + len. this len is then checked against dataset ndims in the handle_dataset function of h5dump.
Changed all references to use new data structure.
Added tests for each subset parameter.
r19190:
Added new h5dump ddl files
Tested on:
Mac OS X/32 10.6.4 (amazon) w/debug & production
(h5committested on Coverity branch)
1. For enum type, members shouldn't be defined yet.
2. H5T_ORDER_NONE only works for reference and fixed-length
string.
3. For opaque type, the order will be ignored.
4. For compound type, all restrictions above apply to the
members.
I'll change H5Tget_order and do another commit.
Tested on jam, heiwa, and amani.
Bring r19247 from trunk to 1.8 branch:
Bring r19246 from Coverity branch to trunk:
Improve metadata "accumulator" for files to track the dirty region within
the accumulator data. This allows the accumulator to drop clean data and
reduces the number of I/O operations when it is dirtied.
This is progress toward Bz#1142 - speeding up the fheap test (which it
doesn't help much, actually, but it does help other aspects of file I/O).
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
(h5committested on trunk)
Bring r19242 from trunk to 1.8 branch:
Add in override for testing timeout alarm to allow "HDF5_ALARM_SECONDS"
environment variable to change the timeout value from the default value.
Tested on:
Mac OS X/32 10.6.4 (amazon) w/debug & production
(Too minor to require h5committest)
Bring r19237 from trunk to 1.8 branch:
Clean up misc. compiler warnings.
Tested on:
FreeBSD/32 6.3 (duty) 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
Initialize loop variable that caused failures in certain circumstances.
Also clean up compiler warnings and release MPI datatype.
Tested on:
Linux/64-amd64 2.6 (abe) w/parallel, w/FORTRAN, in debug mode
Description:
honest3 v1.8 failed in parallel test. It got stuck in the same
testpar/testphdf5 subtest (cbhsssdrpio). This is an old problem.
Upon closer inspection, the testphdf5, when terminated, had clocked
up 1hr 9min 46 sec wall clock time. Honest1 system also sent a message
that an mpi process has used up 30+ CPU minutes which exceeded their login
node cpu time limit and they killed the process. I also did a hand-run
of testphdf5. All subtests before cbhsssdrpio completed in a few minutes.
Therefore, it is safe to say the majority of the 70 minutes of wall clock
time are spent in the sub-test cbhsssdrpio. It also used up lots of CPU
time. cbhsssdrpio is likely infinite looping.
Since MPI application is prone to infinite looping due to message deadlock,
the testphdf5 has a built-in protection to give each subtest at most 20 minutes
of wall-clock time to run. When the 20 minutes wall-clock time is exceeded,
the testphdf5 will attempt to terminate itself. This prevents unnecessary
CPU time consumption in infinite looping.
But that clock limit was changed to 30 and then 60 minutes. I should have
but failed to, noticed the change mentioned by Quincey. IMO, 20 wall clock
time is more than sufficient for each subtest of testphdf5 to complete.
If a subtest takes longer than 20 minutes, it is likely infinite looping.
Giving it more time will not help.
If a subtest of testphdf5 takes more than 20 minutes, it should be broken
down to small tests that will finish way under 20 minutes so that it is
much easier to see progress and identify any deadlock problems.
In view of this, I am changing the testphdf5 time limit back to 20 minutes.
This will at least stop the CPU TIME exceeding limits and annoying the
system administrators.
Maybe there could be a provision, such as environment variable like
$HDF5_ALARM_SECOND to modify the alarm duration on individual execution.
Even so, that should be used temporary to see if an execution just needs
a little more time.
Tested: just eyeballed as the change is trivia.
It is an error to use the condition H5_HAVE_FSEEK64 to control the definition
of HDlseek. It caused errors in AIX where lseek64 is available.
Replaced it with H5_HAVE_LSEEK64. Also added the missing HDstrcasecmp macro.
Tested: AIX using default and --disable-largefile.
fseek64 was used to support large file access for the STDIO driver back in
version 1.2.2 in year 2000. Some how it was not included in version 1.4.0.
Now, fseeko64 is used to support large file. There is no more need for fseek64
which is not a standard call. Removed its presence from configure and related
files.
Tested: jam for configure only.
The STDIO only checked for fseeko and incorrectly assumed it can support file
sizes larger than 32bits. Fixed it by making to use fseeko64 if supported, else
use fseeko. To simplify the code, assume fseeko which is a POSIX function must
be supported. Therefore, fseek is not used at all.
(Note: the above applies to Unix-like system. The Windows platform has
hardcoding using Windows functions which are NOT POSIX compliant.)
Tested: h5committested. Also tested in BP (AIX) 32/64 and enable/disable-largefile.
In some machine (Linux), when --disable-largefile is used, it claims it has
fseeko64 but off64_t is NOT supported. Moved the test of fseeko64 and ftello64
to where fseek64 is so that they are tested only if off64_t is supported.
Tested: h5committested.
Bring r19179 from trunk to 1.8 branch:
Bring changes from Coverity branch back to trunk:
r19079 & 19080:
[BZ1942] h5dump -u to generate XML, it does not respect the -m option
xml version of dump_data function didn't check for use of fp_format variable.
Added new test expected file for committed bug 1942
r19103, 19104 & 19105:
[BZ1821] h5repack -v did not display correct output for a selected compression. Needed new test for comparing output of -v option.
Added new test file for solution to BZ1821
BZ1821 - Bring test changes from the shell script actually used.
Tested on:
Mac OS X/32 10.6.4 (amazon) debug & production
(h5committested on branch)
herr_t. To minimize the change of the library's behavior, in the function
H5Z_prelude_callback of H5Z.c, if the return value of can_apply is FALSE and
the filter is MANDATE, this function returns a FAILURE. If the return value is FALSE
but the filter is OPTIONAL, this function returns a SUCCEED. During the IO, the filter
will fail and return a size of zero. But the pipeline will skip this filter.
Tested on jam, linew, and amani. Tested on jam with szip.
The previous fix had the Windows code in H5private.h but they should have been
in H5win32defs.h which holds all Windows-specific definitions. Moved the fix.
Tested: BP (AIX) to confirm the fix is still valid. Windows tests will occur
in daily tests tonight.
Description:
test/big incorrectly determined not able to write files larger than 2GB and
skipped the SEC2 and STDIO driver tests. The reason was because it was using
off_t while the SEC2 driver is using lseek64 which expects off64_t type.
Solution:
Created a new HDoff_t which is set to off_t or off64_t or other appropriate
type depending on which of lseek or lseek64 is available. Changed SEC2 file
driver and the big test to use this common definition.
Tested:
In BP (AIX), using --enable and --disable-largefile, for both 32 and 64 bits
modes. Did not do h5committest because: 1. the error was exposed in the remote
BP machine; 2. the change is trivial.
