| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bring changes from trunk from the time the branch was created (r14280)
up to the 1.8.0 release (r14525) back into the metadata journaling branch.
Tested on:
FreeBSD/32 6.2 (duty) in debug mode
FreeBSD/64 6.2 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (kagiso) w/PGI compilers, w/C++ & FORTRAN, w/threadsafe,
in debug mode
Linux/64-amd64 2.6 (smirom) w/default API=1.6.x, w/C++ & FORTRAN,
in production mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, in production mode
Linux/64-ia64 2.4 (tg-login3) w/parallel, w/FORTRAN, in production mode
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Convert the symbol table node metadata cache client to use the new
journaling cache callbacks.
Also added a 'H5F_t *' parameter to the 'serialize' callback for the
journaling cache, which makes the client's job much easier.
Various minor coding cleanups, etc. also.
Tested on:
FreeBSD/32 6.2 (duty) in debug mode
FreeBSD/64 6.2 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (kagiso) w/PGI compilers, w/C++ & FORTRAN, w/threadsafe,
in debug mode
Linux/64-amd64 2.6 (smirom) w/default API=1.6.x, w/C++ & FORTRAN,
in production mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, in production mode
Mac OS X/32 10.5.3 (amazon) in debug mode
Linux/64-ia64 2.4 (tg-login3) w/parallel, w/FORTRAN, in production mode
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Updates to test opening of a file created with journaling,
along with associated debugging modifications.
(Mike M. To get journal deletion to work correctly, I
had to modify H5C2_jb__init() to allocate a buffer for the
journal file name and copy it into the buffer. Similarly,
I had to modify to H5C2_jb__takedown() to free the buffer.
The fix was hurried, and should be reviewed. Also, a
similar fix is probably in order for the HDF5 file name.)
* Fix for the bug Albert reported on Linew.
* An attempt to apply the changes Quincey requested to the
loc_id parameters to the FUNC_ENTER_API_META macro calls
in:
H5Gmove2(), (src_loc_id --> dst_loc_id)
H5Lcopy(), (src_loc_id --> dst_loc_id)
H5Lmove(), (src_loc_id --> dst_loc_id)
H5Glink2(), (cur_loc_id --> new_loc_id)
H5Lmove() (cur_loc_id --> new_loc_id)
However, with the exception of the requested change to
H5Gmove2(), all these chages caused us to fail the
regression tests. Thus only the H5Gmove2() change is
made.
Several caviats and warnings:
* If you build and test this checkin, it will fail on the
on the test for trecover.
This showed up after I updated my project, so initially
I thought I had broken something. However, after examining
the problem for a while, I thought to checkout the version
prior to this checkin, and test to see if the problem appeared.
It did (under serial on phoenix, and parallel on Kagiso),
so I am going ahead with this checkin regardless under the
assumption that it is orthoginal to my changes.
* Low level testing for the journaling feature of the metadata
cache is not complete. The coverage of the existing tests
is good enough that I don't expect anything major, but don't
be surprised if you run into problems around the edges.
In particular, enabling and disabling journaling while the
file is open has not been tested at all. Suggest we stay
away from this until it gets at least a once over.
* The metadata journaling smoke check tests in cache2_journal.c
are still configured to generate the architype files used to
check journal output. This can be turned off any time, but
given Quincey's constaints on test file size, I have to write
code to skip the tests if the architype files are missing,
and then put compressed versions of the architype files in
svn before I do so. Unfortunately, there is no time before
I leave.
* I left a good bit of debugging code in both the journaling
code proper, and the associated test code. It should all
be #if 0'ed out at present, but if you run into it, you
know what is going on. Needless to say, I'll delete it when
I finish testing.
* I was not able to reproduce the bug Albert observed on RedStorm
locally, so I don't have a fix for it. That said, I touched
some things that could have caused it, so it is possible that
I fixed it by accident.
Testing:
Before I updated, I was able to build and test serial on Phoenix
and Linew, and parallel on Kagiso without errors in the regression
tests.
As discussed above, after the update, I failed in the test for
trecover in a serial build and test on Phoenix, and parallel build
and test on Kagiso. Linew is slow, so I didn't attempt a test there.
Since the same failures appear in the verion prior to this checkin,
I am going ahead with the checkin regardless on the assumption that
the problem is orthoginal to my changes.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
While this code doesn't break any of the existing tests, it HAS NOT
been tested beyond that.
Also mods needed to integrate the journaling code with Quincey's latest
mods, and to adapt existing test code to slight changes caused by the
addition of journaling.
Finally, fixed an undefined variable bug in the HL code exposed by the
journaling mods.
Tested serial under Linux (Phoenix) and parallel under Linux (Kagiso).
|
|
|
|
|
|
|
|
|
|
| |
Add "_META" suffix to FUNC_ENTER/FUNC_LEAVE API routines that can modify
metadata in the file. This will give us a single place to change to recording
the beginning & ending of transactions.
Tested on:
FreeBSD/32 6.2 (duty)
Too simple to require h5committest
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Switch v1 B-tree nodes from using previous cache to use the new journaling
cache code. This is a major switch for the cache callbacks.
Switched the metadata caching code from having a pointer to the file it's
in to receiving a pointer to the file, when needed. This was necessary in
order to avoid crashing when two files IDs were open on the same underlying
file and one of those files was closed with cache entries using the file
pointers. Also took out the check in the caching code for reading off the
end of the file, which didn't play nicely with the multi-file VFD.
Switching the cache from having a pointer internally to requiring one
externally meant tweaking almost all the test code. :-/
Added correct exit codes to cache2 tests that didn't have them already,
so the 'make check' will stop when they fail.
Use the path to the h5diff in this build in the tools/h5recover testing
script, since we can't guarantee a user has HDF5 already installed.
Assorted minor tweaks to get everything to compile more cleanly and pass
all the tests (on my Mac :-).
Tested on:
Mac OS X (10.5.2) w/C++
(more testing coming up shortly, I just didn't have my "rsync testbed" set
up for this branch when I started making changes to the code)
|
|
|
|
|
|
| |
metadata
journaling branch. This version may not compile.
|
|
|
|
|
|
|
|
|
|
| |
Removed the stream-vfd from the basic library code.
(The stream-vfd source files are not removed yet but the MANIFEST has been
updated to NOT release those stream-vfd source files.)
Platforms tested:
Kagiso and smirom.
Then test the release tar ball in kagiso.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Remove stray reference to "group revision" configure flag
Remove Windows-specific EOL characters from src/H5FDwindows.c
Add some missing types to bin/trace
Re-run bin/reconfigure script to regenerate information after changes above.
Tested on:
FreeBSD/32 6.2 (duty)
FreeBSD/64 6.2 (liberty)
Linux/32 2.6 (linux)
Solaris/32 2.10 (linew)
|
|
|
|
|
|
|
|
|
|
|
| |
Various code cleanups and refactor recent changes for h5stat to fit
into the existing library data structures better.
Tested on:
Mac OS X/32 10.4.10 (amazon)
FreeBSD/32 6.2 (duty)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
New feature.
Description:
Added routines to report on the amount of storage for:
1) 1.6 btree and heap storage info for groups
2) 1.8 btree, fractal heap storage info for groups, attributes and SOHM table
3) btree storage for chunked datasets
4) 1.8 superblock extension size.
Platform tested:
h5committested.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
into Makefile.am un
der /src. So when gmake check-install is run for HDF5 examples, it will fail. Th
e fix is to include a macro #ifdef H5_HAVE_WINDOW around #include H5FDwindows.h
so that H5FDwindows.h will only be included on windows, not on other platforms.
We've fixed with the same way for other source code that includes H5FDwindows.h.
This has been tested with h5committest. However, it takes a few hours to test all tests and except kagiso. Tests on linew and smirom cannot finish. It takes forever for the fheap test.
And for parallel test at kagiso, ph5diff test seems to be hanged. I think all of our changes will only affect windows. So in order not to block other developers to check in their code, we just go ahead to check our changes.
Kent Yang
|
|
|
|
|
|
|
| |
windows as the default vfd. Although sec2 is still default on Windows, it should now be possible to switch by simply changing H5pubconf.h .
Tested:
.NET 2003 on WinXP
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The main purpose of this checkin was to eliminate the
space used for tracking creation time indices when there is no way they
can be used (i.e. attributes can't be shared in the file and the user hasn't
turned on attribute creation tracking), however there were some other minor
changes which crept in:
- Fix a cache locking deadlock when a shared attribute and one of its
components end up in the same fractal heap direct block.
(This is fixed the "slow" way for right now, until John has time
to add support for readers/writer locking to the cache.
- Optimize attribute copying when a copy will be kept during a v2 B-tree
search.
- When freeing a block on disk, attempt to merge it with the metadata
and "small data" aggregators.
Tested on:
Mac OS X/32 10.4.8 (amazon)
FreeBSD/32 6.2 (duty)
|
|
|
|
|
|
|
|
|
|
|
|
| |
running on kagiso.
The static szip libraries are working as intended, but when shared szip
libraries are present and intended for use they cannot be opened. A check
has been implemented to test if shared szip libraries are functional on
current platform. If they are not, and the user is trying to use them,
szip is now disbaled and the user is informed.
This issue occurs on kagiso, and the resolution was tested on kagiso.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix bug in external links when "strong" file close degree is used.
Make external link callbacks use standard error reporting mechanisms.
Other minor code cleanups...
Tested on:
Mac OS X/32 10.4.8 (amazon)
FreeBSD/32 6.2 (duty)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Revise latest form of superblock format pretty drastically, to
eliminate unused fields and move rarely used fields into superblock extension.
Finished removing last vestiges of references to (never used) i"shared"
object header message ID.
Added object header messages for non-default v1 B-tree 'K' values
and for driver info.
Updated testfiles to reflect size changes, etc.
Various minor cleanups, etc.
Tested on:
FreeBSD/32 6.2 (duty)
Mac OS X/32 10.4.8 (amazon)
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Further cleanups to superblock code:
- Move most of the ISOHM-specific code into the H5SM module
- Make only one checksum, for combined superblock & driver info
block
Minor other cleanups, etc.
Tested on:
Mac OS X/32 10.4.8 (amazon)
FreeBSD/32 6.2 (duty)
|
|
|
|
|
|
|
|
|
|
| |
Straighten out some of the code for handling superblocks, in preparation
for revising the format of the latest version of the superblock before the
beta release.
Tested on:
Mac OS X/32 10.4.8 (amazon)
FreeBSD/32 6.2 (duty)
|
|
|
|
|
|
| |
helpful comments.
Tested on Windows, smirom, and kagiso.
|
|
|
|
|
|
|
|
|
| |
copyright notice.
Tested platform:
Kagiso only since it is only a comment block change. If it works in one
machine, it should work in all, I hope. Still need to check the parallel
build on copper.
|
|
|
|
|
|
|
|
|
| |
object header
information, rather than storing it directly in the superblock.
This should make future updates to the superblock much less painful.
Tested on Windows, kagiso, and smirom.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Move datatype messages into new shared message method interface
Clean up various formatting & compiler warnings
Minor optimizations
Tested on:
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
|
|
|
|
|
|
|
| |
The problem
came from the EOA for the whole MULTI file. It's taken out because it's
meaningless for MULTI file. Instead, each individual file has its EOA.
|
|
|
|
|
|
|
|
|
| |
Moved SOHM table version out of table encoding and completely into superblock.
This is a file format change.
Added test that extends shared dataspaces. Dynamically allocate arrays in
shared message cache code. Clean up comments.
Tested on windows, kagiso, smirom.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add support for deleting all the attributes on an object, when they
are dense and possibly shared.
Also, add some testing routines, for better error detection.
Should fix daily test failures also...
Tested on:
FreeBSD/32 6.1 (duty)
|
|
|
|
|
|
|
|
|
| |
haven't
been run in a while, so many of the source files were updated with
tracing macros, etc. No code changes by me.
Tested on kagiso and smirom.
|
|
|
|
|
|
|
|
|
|
|
| |
Migrate more object header routines to use the H5O_msg_ prefix and put
them into the src/H5Omessage.c code module.
Tested on:
Mac OS X/32 10.4.8 (amazon)
FreeBSD/32 4.11 (sleipnir)
Linux/32 2.4 (heping)
AIX/32 5.? (copper)
|
|
|
|
|
|
|
|
|
|
|
| |
SOHM table
version and size information into the superblock to eliminate a read when
loading it.
This is a file format change, and hopefully the last one (knock on wood).
Tested on kagiso and Windows (mostly just a SOHM change).
|
|
|
|
|
|
|
|
|
|
|
|
| |
This feature is still in progress; Shared Object Header Messages are not
complete as a feature and are not thoroughly tested. There are still
"TODO" comments in the code (comments with the word "JAMES" in them,
so as not to be confused with other TODO comments).
Hopefully this checkin will reduce the liklihood of conflicts as I finish
implementing this feature.
All current tests pass on juniper, copper (parallel), heping, kagiso, and mir.
|
|
|
|
|
|
|
| |
Clean up initialization of H5F & H5L APIs.
Tested on:
Linux/64 2.6 (chicago2)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Refactor generic property list initialization code to put property list
specific routines in property list modules, instead of scattered to the four
winds. Also, introduce property list class initialization objects, to make
adding new property list classes in the library easier.
Fix daily test failure by using H5Pget_elink_prefix() API routine instead
of looking at the "raw" generic property list information.
Tested on:
Mac OS X/32 10.4.8 (amazon)
FreeBSD/32 4.11 (sleipnir) w/threadsafe
Linux/32 2.4 (heping) w/C++ & FORTRAN
Linux/64 2.4 (mir) w/build-all & 1.6 compat
|
|
|
|
|
|
|
|
|
|
| |
Altix (cobalt). There's a configure
option --enable-direct-vfd/--disable-direct-vfd to enable/disable Direct I/O support. The default
is enabled. There's a small test in test/vfd.c. Another way to test it is to set environment
variable HDF5_DRIVER to "direct" and run "make check" in the test/ directory. There'll be some
further improvement in the following checkin including allowing user to provide memory boundary
value, file block size, and copying buffer size.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
another bug
exposed by that one in which closing a read/write file ID wouldn't flush the
cache if there was another file handle open. This caused problems when the
second file handle was read-only; neither would actually flush the cache.
Now all read/write file handles flush the cache when the close, but only
when the file is going to be destroyed is the cache destroyed. This way
read-only file handles keep the cache open but it is always clean.
Tested on mir, heping, sol, and copper (parallel)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add the "use the latest version of the format" support to datatype
messages. And a regression test to check that it's working.
Also, found that we have been over-allocating space for compound datatype
messages (when they have array datatypes or can use the latest version of the
format) and trimmed the size back.
Clean up datatype & dataspace encode/decode routines by having them
allocate & release "fake" file structures, which gets them out of needing to
"know" about the internals of file structures.
Other minor whitespace/formatting cleanups, etc.
Tested on:
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix file handle destroy routine to not attempt to flush out partially
initialized file handles (when opening a file fails).
Tested on: (until they finished testing, this time... :-)
FreeBSD/32 4.11 (sleipnir) w/threadsafe
Linux/64 2.4 (mir) w/1.6 compat
Linux/32 2.4 (heping) w/FORTRAN & C++
Mac OSX/32 10.4.8 (amazon)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add test to fractal heaps to exercise issues with opening the same heap
through two different file handles.
Fix issues with file handle contexts in metadata cache callbacks for heap
components.
Fix bug in file close handling where cached information was being
invalidated even when another file handle was open to the file.
Tested on:
FreeBSD/32 4.11 (sleipnir) w/threadsafe
Linux/64 2.4 (mir) w/1.6 compat
Linux/32 2.4 (heping) w/FORTRAN & C++
Mac OSX/32 10.4.8 (amazon)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
File format is not stable, don't keep files produced!
Description:
First stage of checkins modifying the format of groups to support creation
order. Implement "dense" storage for links in groups.
Try to clarify some of the symbols for the H5L API.
Add the H5Pset_latest_format() flag for FAPLs, to choose to use the newest
file format options (including "dense" link storage in groups)
Add the H5Pset_track_creation_order() flag for GCPLs, to enable creation
order tracking in groups (although no index on creation order yet).
Remove --enable-group-revision configure flag, as file format issues are
now handled in a backwardly/forwardly compatible way.
Clean up lots of compiler warnings and other minor formatting issues.
Tested on:
FreeBSD/32 4.11 (sleipnir) w/threadsafe
Linux/32 2.4 (heping) w/FORTRAN & C++
Linux/64 2.4 (mir) w/enable-v1.6 compa
Mac OSX/32 10.4.8 (amazon)
AIX 5.3 (copper) w/parallel & FORTRAN
|
|
|
|
|
|
|
|
|
| |
Review, revise & checkin in Peter's latest round of object copy changes,
which add basic support for datasets & attributes with reference datatypes.
Tested on:
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
|
|
|
|
|
|
|
| |
Re-run 'bin/reconfigure' script after recent checkins
Tested on:
none - shouldn't have any affect on compilation
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add "use the latest version of the file format" flag to the file access
property list and internal file data structures.
Fix bug where metadata block size was retrieved instead of the small
data block size.
Categorize property list routine prototypes in the public header file.
Tested on:
Mac OS/PPC 10.4 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
|
|
|
|
| |
correctly.
|
|
|
|
|
|
|
|
|
| |
(read/write or
read-only).
Added this to external links, so that external files are opened with the
same intent as the source file.
Added tests.
|
|
|
|
|
|
|
|
|
|
|
| |
Refactored free space manager to use metadata cache for serialized free
space sections. This speeds up the fractal heap test considerably...
Tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4/32 (chicago)
Linux 2.4/64 (mir)
Mac OS X (amazon)
|
|
|
|
|
|
|
|
|
|
|
| |
Code cleanup
Description:
Trim trailing whitespace in Makefile.am and C/C++ source files to make
diffing changes easier.
Platforms tested:
None necessary, whitespace only change
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Code checkpoint
Description:
Checkpoint fractal heap improvements, as well as move the free space
manager code that it's using into a separate package.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4/64 (mir) w/C++ & FORTRAN
Linux 2.4/32 (heping)
Solaris 2.9 (shanti)
AIX 5.? (copper) w/FORTRAN & parallel
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Code maintenance
Description:
Remove flexible parallel code
Platforms tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4 (heping)
Solaris 2.9 (shanti)
Linux 2.4/64 (mir)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
New feature
Description:
Check in baseline for compact group revisions, which radically revises the
source code for managing groups and object headers.
WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!!
WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!!
This initiates the "unstable" phase of the 1.7.x branch, leading up
to the 1.8.0 release. Please test this code, but do _NOT_ keep files created
with it - the format will change again before the release and you will not
be able to read your old files!!!
WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!!
WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!! WARNING!!!!
Solution:
There's too many changes to really describe them all, but some of them
include:
- Stop abusing the H5G_entry_t structure and split it into two separate
structures for non-symbol table node use within the library: H5O_loc_t
for object locations in a file and H5G_name_t to store the path to
an opened object. H5G_entry_t is now only used for storing symbol
table entries on disk.
- Retire H5G_namei() in favor of a more general mechanism for traversing
group paths and issuing callbacks on objects located. This gets us out
of the business of hacking H5G_namei() for new features, generally.
- Revised H5O* routines to take a H5O_loc_t instead of H5G_entry_t
- Lots more...
Platforms tested:
h5committested and maybe another dozen configurations.... :-)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug fix
Description:
Fix "double root open" bug and apply same fix to datasets & named datatypes.
Also fix bug in symbol table entry lookup code that would "transfer" an
object from one file handle to another file handle when the two file handles
were opened on the same actual file.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Too minor to require h5committest
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Repair synchronization bug in the metadata cache in PHDF5
Also repair numerous other bugs that surfaced in testing the
bug fix.
Description:
While operations modifying metadata must be collective, we allow
independant reads. This allows metadata caches on different processes
to adjust to different sizes, and to place the entries on their dirty
lists in different orders. Since only process 0 actually writes
metadata to disk (all other processes thought they did, but the writes
were discarded on the theory that they had to be collective), this made
it possible for another process to modify metadata, flush it, and then
read it back in in its original form (pre-modification) form. The
possibilities for file corruption should be obvious.
Solution:
Make the policy that only process 0 can write to file explicit, and
visible to the metadata caches. Thus only process 0 may flush dirty
entries -- all other caches must retain dirty entries until they are
informed by process 0 that the entries are clean.
Synchronization is handled by counting the bytes of dirty cache entries
created, and then synching up between the caches whenever the sum
exceeds an (eventually user specified) limit. Dirty metadata creation
is consistent across all processes because all operations modifying
metadata must be collective.
This change uncovered may bugs which are repaired in this checkin.
It also required modification of H5HL and H5O to allocate file space
on insertion rather than on flush from cache.
Platforms tested:
H5committest, heping(parallel & serial)
Misc. update:
|