Clean up ISOHM code further and get rid of several non-optimal ways
of working with object headers.
Tested on:
FreeBSD/32 6.2 (duty)
Mac OS X/32 10.4.9 (amazon)
Check in "unique, but sharable" optimization to ISOHM code, which
allows object header messages that are only used in one object to remain in
the sole user's header, but migrates messages that are used in more than one
header into the ISOHM heap.
Tested on:
Mac OS X/32 10.4.9 (amazon)
FreeBSD/32 6.2 (duty)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Fix H5O_msg_iterate() and H5O_link() to protect cache entries with
better permissions.
Tested on:
Mac OS X/32 10.4.9 (amazon)
FreeBSD/32 6.2 (duty)
FreeBSD/64 6.2 (liberty)
Take further advantage of the readers/writer locking in the metadata cache
and push the locking of the local heap up to the routines which call B-tree
routines, which should drastically lower the number of metadata cache protect/
unprotect calls.
Tested on:
Mac OS X/32 10.4.9 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Initial changes to take advantage of readers/writer locking on metadata
cache entries. Reduce the # of protect/unprotect pairs during B-tree iteration
by holding reader lock on B-tree and symbol table nodes while iteration occurs.
This also has the benefit of preventing an iteration callback from
modifying the B-tree being iterated over (which would probably confuse the
B-tree iteration code terribly)
Tested on:
Mac OS X/32 10.4.9 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Don't delete chunks from the dataset's B-tree while we are iterating over
the B-tree, when reducing the size of the dataset's dataspace with
H5Dset_extent().
Tested on:
Mac OS X/32 2.6 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
Add "fail if unknown" and "mark if unknown" flags to object header messages.
This gives the library a clean way to tell older libraries that a message should
not be just ignored if it's unknown, but that other actions should occur.
Tested on:
Mac OS X/32 10.4.9 (amazon)
FreeBSD/32 6.2 (duty)
FreeBSD/64 6.2 (liberty)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
files. 2. added a check in test/big.c to make sure the system supports big file.
Tested on smirom because already tested the same changes for v1.6 on liberty, smirom,
copper, and sol.
Fully enable readers/writer locking in metadata cache and correct errors
in cache clients which were modifying a cache entry while only holding a
read lock on it.
Tested on:
Mac OS X/32 10.4.9 (amazon)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
FreeBSD/32 6.2 (duty)
FreeBSD/64 6.2 (liberty)
H5Zpublic.h file didn't include H5public.h; done
H5Spublic.h had ghost function H5Sset_space; deleted
New source files were added to make.com for VMS.
h5gget_comment_c had a small bug in calculating buffer length; fixed
Platforms tested: pahra and kagiso; too minor for h5committest
Rename new error handling API routines from H5E<foo>_stack() to
H5E<foo>2().
Tested on:
Mac OS X/32 10.4.9 (amazon)
FreeBSD/32 6.2 (duty)
FreeBSD/64 6.2 (liberty)
Change H5[D|G|T]<foo>_expand() "temporary" API routines to
H5[D|G|T]<foo>2() "versioned" routines. Also added
H5[D|G|T](create|commit)_anon() routines to continue to allow "anonymous"
objects to be created in a file.
Tested on:
Mac OS X/32 10.4.9 (amazon)
FreeBSD/32 6.2 (duty)
FreeBSD/64 6.2 (liberty)
Linux/32 2.6 (chicago)
Linux/64 2.6 (chicago2)
cache entries. Added test code to test the new feature. Also some
minor cleanum
h5committested -- passed on copper and sol. Failed on osage with
a configuration error that appears unrelated to my changes. Serial
test (debug mode) passes on Phoenix (Linux x86 2.6 kernel), so I
went ahead with the checkin.
4294967296LL is not supported by older C like Windows VC6.
Solution:
/* use <= 2**32 -1 to stay within 4 bytes integer range */
HDassert(chunk0_size <= 4294967295UL);
Platforms tested:
kelgia, copper and osage.
