platform/upstream/lmdb.git
11 years agoITS#7652 fix I/O error checks
Howard Chu [Tue, 30 Jul 2013 17:22:12 +0000 (10:22 -0700)]
ITS#7652 fix I/O error checks

partial revert of d6d2638acc245116b8f091ac425b6700d06c4713 and
26a25df5fcc2fcddae6597a61c1b867fc27c568b

The original code was already tested and working correctly.

11 years agoTweak mdb_envinfo numreaders
Howard Chu [Mon, 29 Jul 2013 00:02:51 +0000 (17:02 -0700)]
Tweak mdb_envinfo numreaders

Return the actual shared reader count when it exists, not
just the current process env's reader count.

11 years agoITS#7615 use shorter names for semaphores
Howard Chu [Fri, 26 Jul 2013 17:19:54 +0000 (10:19 -0700)]
ITS#7615 use shorter names for semaphores

NetBSD can only handle up to 14 chars, we were using 21. Now
we encode to 15, and for NetBSD truncate the last char.

11 years agoTweak reader_pid check
Howard Chu [Fri, 19 Jul 2013 16:57:33 +0000 (09:57 -0700)]
Tweak reader_pid check

Check again after acquiring rmutex. Avoids potential issue with
a duplicate pid coming in between initial check and rmutex.

11 years agoGet pid lock outside of rmutex
Howard Chu [Fri, 19 Jul 2013 16:55:10 +0000 (09:55 -0700)]
Get pid lock outside of rmutex

Avoid holding rmutex for longer than necessary.

11 years agoTweak reader checks
Howard Chu [Thu, 18 Jul 2013 22:24:09 +0000 (15:24 -0700)]
Tweak reader checks

Use mti_numreaders for loop limit, not me_maxreaders.

11 years agoAdd mdb_reader_check()
Howard Chu [Thu, 18 Jul 2013 17:40:21 +0000 (10:40 -0700)]
Add mdb_reader_check()

11 years agoSplit MDB_VERSION to MDB_DATA/MDB_LOCK VERSION
Howard Chu [Thu, 18 Jul 2013 16:11:09 +0000 (09:11 -0700)]
Split MDB_VERSION to MDB_DATA/MDB_LOCK VERSION

11 years agoTweak reader_list
Howard Chu [Thu, 18 Jul 2013 16:00:51 +0000 (09:00 -0700)]
Tweak reader_list

11 years agoTweak mdb_stat(1)
Howard Chu [Thu, 18 Jul 2013 15:33:24 +0000 (08:33 -0700)]
Tweak mdb_stat(1)

Don't obtain reader txn before displaying reader table. Exit
after reader table if no other DB query options were given.

11 years agoAdd mdb_reader_list()
Howard Chu [Thu, 18 Jul 2013 14:41:11 +0000 (07:41 -0700)]
Add mdb_reader_list()

Dump the active slots in the reader table.

11 years agoAdd mdb_dbi_flags()
Howard Chu [Mon, 15 Jul 2013 17:57:13 +0000 (10:57 -0700)]
Add mdb_dbi_flags()

Retrieve the flags from a DB handle.

11 years agoFix child txn dirty_room counts in spill/unspill
Howard Chu [Sun, 14 Jul 2013 23:53:04 +0000 (16:53 -0700)]
Fix child txn dirty_room counts in spill/unspill

Don't count pages twice if they're already accounted in an ancestor txn.

11 years agoMore for stale sub-cursor flags
Howard Chu [Sun, 14 Jul 2013 15:28:26 +0000 (08:28 -0700)]
More for stale sub-cursor flags

Same fix for cursor_first/last.

11 years agoFix stale sub-cursor C_INIT flag
Howard Chu [Sun, 14 Jul 2013 15:20:18 +0000 (08:20 -0700)]
Fix stale sub-cursor C_INIT flag

Whenever we enter cursor_set() the sub-cursor's flag must be
cleared. If the new cursor position has valid subdata it will
be initialized again, if not then the sub-cursor has nothing
to point to.

11 years agoTweak comments, defaults should be OK already
Howard Chu [Fri, 12 Jul 2013 20:55:18 +0000 (13:55 -0700)]
Tweak comments, defaults should be OK already

11 years agoBump version to 0.9.7
Howard Chu [Fri, 12 Jul 2013 20:36:05 +0000 (13:36 -0700)]
Bump version to 0.9.7

11 years agoAlso set/clear P_KEEP in parent txn's cursors
Hallvard Furuseth [Fri, 12 Jul 2013 09:30:33 +0000 (11:30 +0200)]
Also set/clear P_KEEP in parent txn's cursors

11 years agoSpill pages, take 3
Howard Chu [Thu, 11 Jul 2013 20:09:47 +0000 (22:09 +0200)]
Spill pages, take 3

11 years agoDelay touching pages until cursor is positioned.
Howard Chu [Thu, 11 Jul 2013 20:09:47 +0000 (22:09 +0200)]
Delay touching pages until cursor is positioned.

This avoids unnecessary rewrites of pages that do not change.
(Restructuring for upcoming mdb_page_spill work.)

11 years agoSimplify: Always set C_UNTRACK for tracked cursors.
Hallvard Furuseth [Thu, 11 Jul 2013 20:09:46 +0000 (22:09 +0200)]
Simplify: Always set C_UNTRACK for tracked cursors.

TODO: Rename C_UNTRACK to C_TRACKED.  Omitted now for readability.
The current name is because it's lazy: not always set when tracked.

11 years agoSave freelist using proper mdb_cursor_put().
Hallvard Furuseth [Thu, 11 Jul 2013 20:09:46 +0000 (22:09 +0200)]
Save freelist using proper mdb_cursor_put().

(Restructuring for upcoming mdb_page_spill work.)
mdb_freelist_save() can't just Get() the destination, since
mdb_page_spill() may have put the destination in the read-only map.

TODO: Can this new put() modify the freelist, which would break it? The
final iteration's put() can shorten the node, the rest uses MDB_CURRENT.

We could set P_KEEP on dirty freeDB leaves and ovpages, since they are
all about to be modified.  But the code in this commit must stay anyway,
if mdb should support dropping a 256G DB. I.e. too big for dirty_list.

11 years agoMove code out to mdb_page_dirty()
Howard Chu [Thu, 11 Jul 2013 20:09:46 +0000 (22:09 +0200)]
Move code out to mdb_page_dirty()

11 years agoFactor out parent
Howard Chu [Thu, 11 Jul 2013 20:09:46 +0000 (22:09 +0200)]
Factor out parent

11 years agoFix env_read_header() on Windows
Howard Chu [Fri, 12 Jul 2013 19:53:35 +0000 (12:53 -0700)]
Fix env_read_header() on Windows

Commit d6d2638acc245116b8f091ac425b6700d06c4713 broke read
on zero-length files.

11 years agoDo not follow uninited cursors' page pointers.
Hallvard Furuseth [Wed, 10 Jul 2013 20:11:44 +0000 (22:11 +0200)]
Do not follow uninited cursors' page pointers.

Nor uninited cursors' subcursors' page pointers.

11 years agoFix rebalance/cursor adjust
Howard Chu [Wed, 10 Jul 2013 18:03:51 +0000 (11:03 -0700)]
Fix rebalance/cursor adjust

When collapsing root, must also move cursor index down,
not just the page pointer.

Also in mtest, break from NEXT loops on error, otherwise it just
prints the previous key/data again, which looks confusing.

11 years agoCursors: Clear C_EOF when clearing C_INITIALIZED
Howard Chu [Wed, 10 Jul 2013 15:49:29 +0000 (08:49 -0700)]
Cursors: Clear C_EOF when clearing C_INITIALIZED

11 years agoFixup other cursors after delete op
Howard Chu [Tue, 9 Jul 2013 21:21:35 +0000 (14:21 -0700)]
Fixup other cursors after delete op

11 years agoITS#7515 Reject conflicting page versions.
Hallvard Furuseth [Sun, 7 Jul 2013 15:14:38 +0000 (17:14 +0200)]
ITS#7515 Reject conflicting page versions.

If mdb_page_touch() sees a page in txn's dirty_list, that
is the page version txn's cursors should have.  Fail if
the user may be seeing and depending on another version.

11 years agoITS#7515 Fix tracking of parent txn's cursors.
Hallvard Furuseth [Sun, 7 Jul 2013 15:13:27 +0000 (17:13 +0200)]
ITS#7515 Fix tracking of parent txn's cursors.

Restore mc_flags and xcursors, they were tracked but not merged.
Simplify: Track parent txn's original cursors after backing them
up, instead of tracking copies and merging them back at commit.

11 years agoSimplify MDB_cursor: Drop flags C_ALLOCD,C_SHADOW.
Hallvard Furuseth [Sun, 7 Jul 2013 15:13:27 +0000 (17:13 +0200)]
Simplify MDB_cursor: Drop flags C_ALLOCD,C_SHADOW.

11 years agoSilence more uninit warnings
Hallvard Furuseth [Sat, 6 Jul 2013 19:42:45 +0000 (21:42 +0200)]
Silence more uninit warnings

11 years agoSilence uninit warning in prev commit
Howard Chu [Tue, 2 Jul 2013 09:23:49 +0000 (02:23 -0700)]
Silence uninit warning in prev commit

11 years agoTweaks for MDB_MULTIPLE
Howard Chu [Tue, 2 Jul 2013 09:19:17 +0000 (02:19 -0700)]
Tweaks for MDB_MULTIPLE

Terminate loop on intermediate failures, return count of written items,
document usage.

11 years agoITS#7635 simplify 9474c1a0b62cad57b62855d09a201670385caa80
Howard Chu [Mon, 1 Jul 2013 21:53:29 +0000 (14:53 -0700)]
ITS#7635 simplify 9474c1a0b62cad57b62855d09a201670385caa80

11 years agoITS#7635 fix read txn potential data race
Howard Chu [Mon, 1 Jul 2013 20:41:23 +0000 (13:41 -0700)]
ITS#7635 fix read txn potential data race

11 years agoFix uninit warnings, lseek usage
Howard Chu [Sun, 30 Jun 2013 14:40:02 +0000 (07:40 -0700)]
Fix uninit warnings, lseek usage

11 years agoFix alloc/free issues.
Hallvard Furuseth [Wed, 26 Jun 2013 16:02:52 +0000 (18:02 +0200)]
Fix alloc/free issues.

Page leak, mdb_page_alloc(). On error, don't shorten me_pghead.
Memleak, mdb_ovpage_free(). Free page or keep it in dirty_list.
Bad MIDL, mdb_midl_need(). Fix midl[-1] (allocated size).

11 years agoFactor out some vars, simplify.
Hallvard Furuseth [Wed, 26 Jun 2013 16:02:48 +0000 (18:02 +0200)]
Factor out some vars, simplify.

11 years agoMakefile/user-macro comments.
Hallvard Furuseth [Wed, 26 Jun 2013 16:02:26 +0000 (18:02 +0200)]
Makefile/user-macro comments.

11 years agoTweak I/O, fix last commit.
Hallvard Furuseth [Wed, 26 Jun 2013 16:02:17 +0000 (18:02 +0200)]
Tweak I/O, fix last commit.

11 years agoImprove MDB error handling, drop seek calls.
Hallvard Furuseth [Sat, 22 Jun 2013 21:15:10 +0000 (23:15 +0200)]
Improve MDB error handling, drop seek calls.

Catch I/O errors. Do nothing between OS call failure and ErrCode().
Do not use errno after non-OS-errors like write() >= 0, which could
give a failure return of success (errno 0) or some irrelevant error
code.  Drop seek calls, use pwrite/pread/Windows OVERLAPPED offset.

11 years agoFix Windows I/O.
Hallvard Furuseth [Sat, 22 Jun 2013 21:01:30 +0000 (23:01 +0200)]
Fix Windows I/O.

Don't put a 64-bit filesize in a 32-bit int before shifting
down.  Always pass &sizehi to SetFilePointer->maxsize, so
sizelo not is treated a signed distance.  Hide unused vars
when _WIN32.  Reinitialize OVERLAPPED before reuse.

11 years agoCatch more MDB errors. DPRINTF in mdb_env_reset0.
Hallvard Furuseth [Sat, 22 Jun 2013 20:17:41 +0000 (22:17 +0200)]
Catch more MDB errors. DPRINTF in mdb_env_reset0.

11 years agoTweak MIDLs, catch errors.
Hallvard Furuseth [Sat, 22 Jun 2013 20:10:43 +0000 (22:10 +0200)]
Tweak MIDLs, catch errors.

Grow midls earlier in order to catch errors earlier.  Use
mdb_midl_need() instead of mdb_midl_grow(), then mdb_midl_xappend()
needs no error checks. Factor out mdb_midl_append_range().

11 years agoFactor out MDB variables/expressions, cleanup.
Hallvard Furuseth [Sat, 22 Jun 2013 10:30:04 +0000 (12:30 +0200)]
Factor out MDB variables/expressions, cleanup.

mdb_page_malloc(): Take a txn arg instead of a cursor.

11 years agoRearrange MDB dirty page code.
Hallvard Furuseth [Sat, 22 Jun 2013 09:56:04 +0000 (11:56 +0200)]
Rearrange MDB dirty page code.

Split out mdb_dpage_free(), mdb_page_flush() and clean up.

11 years agoSimplify mdb_page_alloc().
Hallvard Furuseth [Thu, 20 Jun 2013 05:41:35 +0000 (07:41 +0200)]
Simplify mdb_page_alloc().

Merge if() branches. Restore retry=500 when MDB_PARANOID, for clarity.

11 years agoITS#7620: Keep empty IDLs. Tweak mdb_page_alloc().
Hallvard Furuseth [Thu, 20 Jun 2013 05:41:35 +0000 (07:41 +0200)]
ITS#7620: Keep empty IDLs. Tweak mdb_page_alloc().

MDB_env.me_pghead: Don't free it when empty. mdb_ovpage_free()
needs it, but cannot allocate it.
mdb_midl_alloc(): Fill in length=0.
mdb_page_alloc(): Also Skip freeDB if txnid<3, instead of <4,
and consistently DPRINTF consumed IDLs.

11 years agoITS#7623 Clear P_SUBP on conversion from fake page
Howard Chu [Mon, 17 Jun 2013 20:26:11 +0000 (22:26 +0200)]
ITS#7623 Clear P_SUBP on conversion from fake page

11 years agoITS#7515 Nested MDB txns: Inherit txn flags.
Hallvard Furuseth [Thu, 13 Jun 2013 06:58:25 +0000 (08:58 +0200)]
ITS#7515 Nested MDB txns: Inherit txn flags.

Committing a nested txn lost the MDB_TXN_DIRTY flag
in the parent, unless the child had set it too.

11 years agoClean up mdb_page_touch(), mdb_page_copy().
Hallvard Furuseth [Thu, 13 Jun 2013 06:58:24 +0000 (08:58 +0200)]
Clean up mdb_page_touch(), mdb_page_copy().

When copying, round up/down to aligned sizes. Skip the unused portion,
this was not done when touching a page dirty in the parent txn.
No other change in behavior.

Simplify mdb_page_touch(), including: Drop test m3==mc, the condition
is caught below. Don't "modify" the parent's pgno into the same pgno,
when a nested txn copies a parent's page into its freelist.

11 years agoITS#7594 Fix MDB cursor tracking with subDBs.
Hallvard Furuseth [Thu, 13 Jun 2013 06:58:24 +0000 (08:58 +0200)]
ITS#7594 Fix MDB cursor tracking with subDBs.

The tracking code should not change the current cursor.
It did when that was a C_SUB cursor, which should not be
checked against the tracked cursors but their xcursors.

However, do not bother to skip the tracking code for the
current cursor when it would not change that cursor anyway.

11 years agoITS#7594 Invalidate a dropped MDB DB's cursors.
Hallvard Furuseth [Thu, 13 Jun 2013 06:58:24 +0000 (08:58 +0200)]
ITS#7594 Invalidate a dropped MDB DB's cursors.

11 years agoDon't #define _GNU_SOURCE if already defined.
Hallvard Furuseth [Thu, 13 Jun 2013 06:58:24 +0000 (08:58 +0200)]
Don't #define _GNU_SOURCE if already defined.

11 years agoMore for ITS#7620 Fix mdb_ovpage_free().
Hallvard Furuseth [Thu, 13 Jun 2013 06:25:25 +0000 (08:25 +0200)]
More for ITS#7620 Fix mdb_ovpage_free().

Do not binary-search dirty_list, it is unsorted when MDB_WRITEMAP.
Catch errors. In nested txns, put the page in mt_free_pgs after
all since pages dirty in a parent txn would add complexities.

11 years agoPartial revert c2cac4588a40480c020d320b544bc5f8e72adb11
Howard Chu [Wed, 12 Jun 2013 15:41:32 +0000 (08:41 -0700)]
Partial revert c2cac4588a40480c020d320b544bc5f8e72adb11

MDB_NEXT was fine before, duh.

11 years agoDrop me_pgfree, add mdb_freelist_save().
Hallvard Furuseth [Wed, 12 Jun 2013 15:20:42 +0000 (17:20 +0200)]
Drop me_pgfree, add mdb_freelist_save().

Split up saving me_pghead, to make me_pgfree unneeded. Also mf_pghead
is now a midl. Needed after e7f6767ea815fe0ada1f95037dfdec176ec4d5bb
("Return fresh overflow pages to current pghead").
Tweak MDB_DEBUG freelist output, make it ascending.

11 years agoFix CURSOR_NEXT/PREV on emptied DB
Howard Chu [Wed, 12 Jun 2013 00:13:08 +0000 (17:13 -0700)]
Fix CURSOR_NEXT/PREV on emptied DB

11 years agoMake sure mdb_stat() gets valid data
Howard Chu [Sat, 8 Jun 2013 21:10:08 +0000 (14:10 -0700)]
Make sure mdb_stat() gets valid data

11 years agoReturn fresh overflow pages to current pghead
Howard Chu [Sun, 5 May 2013 08:28:12 +0000 (01:28 -0700)]
Return fresh overflow pages to current pghead

And remove them from the current dirty list.

11 years agoITS#7594 more for subDB cursor fix
Howard Chu [Wed, 5 Jun 2013 23:13:43 +0000 (16:13 -0700)]
ITS#7594 more for subDB cursor fix

11 years agoITS#7594 better fix
Howard Chu [Wed, 5 Jun 2013 22:23:54 +0000 (15:23 -0700)]
ITS#7594 better fix

Update the subDB cursor, don't invalidate it

11 years agotweak mdb_copy, trap signals
Howard Chu [Thu, 30 May 2013 22:56:30 +0000 (15:56 -0700)]
tweak mdb_copy, trap signals

11 years agoWindows portability fixes for prev commit
Howard Chu [Thu, 30 May 2013 22:33:59 +0000 (15:33 -0700)]
Windows portability fixes for prev commit

11 years agoAdd warning about interrupting copy
Howard Chu [Thu, 30 May 2013 20:13:33 +0000 (13:13 -0700)]
Add warning about interrupting copy

11 years agoFix prev commit
Howard Chu [Thu, 30 May 2013 20:09:28 +0000 (13:09 -0700)]
Fix prev commit

11 years agoAdd mdb_env_copyfd()
Howard Chu [Thu, 30 May 2013 20:06:12 +0000 (13:06 -0700)]
Add mdb_env_copyfd()

Allow writing backup to an already opened file handle, for piping
to tar/gzip/ssh/whatever.

11 years agoAdd _M_IX86 macro for MSVC
Howard Chu [Sat, 25 May 2013 17:16:55 +0000 (10:16 -0700)]
Add _M_IX86 macro for MSVC

11 years agoITS#7594 De-init other subcursors in page_touch
Howard Chu [Thu, 23 May 2013 15:13:08 +0000 (08:13 -0700)]
ITS#7594 De-init other subcursors in page_touch

11 years agoDrop unused liblmdb MIDL-range support.
Hallvard Furuseth [Tue, 21 May 2013 21:58:57 +0000 (23:58 +0200)]
Drop unused liblmdb MIDL-range support.

11 years agoFactor out mdb_find_oldest,mdb_dlist_free,dirty_list.
Hallvard Furuseth [Tue, 21 May 2013 21:55:13 +0000 (23:55 +0200)]
Factor out mdb_find_oldest,mdb_dlist_free,dirty_list.

Do not rescan reader table (mdb_find_oldest) after "goto again".
Skip clearing dirty_list[nonzero].mid in mdb_dlist_free(); it
was not done in mdb_reset0() anyway.

11 years agomdb_stat cleanup.
Hallvard Furuseth [Tue, 21 May 2013 21:48:27 +0000 (23:48 +0200)]
mdb_stat cleanup.

Exit with success when there was no failure.
Do not use data containing NUL as a DB name (which is a C string).

11 years agoITS#7598 Tweak MDB_<NEXT/PREV>_NODUP,fix mdb_stat.
Hallvard Furuseth [Tue, 21 May 2013 20:44:51 +0000 (22:44 +0200)]
ITS#7598 Tweak MDB_<NEXT/PREV>_NODUP,fix mdb_stat.

MDB_NEXT_NODUP, MDB_PREV_NODUP: Allow for non-MDB_DUPSORT databases.
No mdb.c code changes needed.

mdb_stat.c: Use MDB_NEXT_NODUP, to avoid a crash with a DUPSORT mainDB.

11 years agoITS#7598 mdb_dbi_open(named DB): Check mainDB flags.
Hallvard Furuseth [Tue, 21 May 2013 17:04:52 +0000 (19:04 +0200)]
ITS#7598 mdb_dbi_open(named DB): Check mainDB flags.

Reject attempts to open named databases if the main
database has flag MDB_DUPSORT or MDB_INTEGERKEY.

DUPSORT would require an xcursor for the DB, INTEGERKEY
would expect the DB name to be a binary integer.

11 years agoITS#7515 Fix nested txn touch of subpage/ovpage.
Hallvard Furuseth [Mon, 20 May 2013 21:12:52 +0000 (23:12 +0200)]
ITS#7515 Fix nested txn touch of subpage/ovpage.

mdb_page_touch(): Don't touch a subpage, replacing with non-subpage.
mdb_cursor_put(): Don't overwrite ancestor txn's dirty overflow page.

11 years agoTweak mdb_page_malloc(),mdb_page_get() semantics.
Hallvard Furuseth [Mon, 20 May 2013 21:12:26 +0000 (23:12 +0200)]
Tweak mdb_page_malloc(),mdb_page_get() semantics.

mdb_page_malloc(): Add "number of pages" parameter.
mdb_page_get(): Add output param for how page was found.
Do not set return params on error.
mdb_cursor_put(): Catch mdb_page_get() error.

Prepares for next commit, no change in caller behavior other
than on mdb_page_get error.

11 years agoMore ITS#7589 followup: OVPAGES() -> mp_pages.
Hallvard Furuseth [Sun, 19 May 2013 17:08:07 +0000 (19:08 +0200)]
More ITS#7589 followup: OVPAGES() -> mp_pages.

11 years agoMore ITS#7589: Fix prev commit.
Hallvard Furuseth [Sun, 19 May 2013 16:38:13 +0000 (18:38 +0200)]
More ITS#7589: Fix prev commit.

mp_pages gives #pages in the ovpage. OVPAGES() no longer does.

11 years agoITS#7589 mdb_cursor_put(): Update ovpage nodesize.
Hallvard Furuseth [Fri, 17 May 2013 17:31:17 +0000 (19:31 +0200)]
ITS#7589 mdb_cursor_put(): Update ovpage nodesize.

Update the nodesize when overwriting an overflow page.
As before, do not attempt to shrink the page.

11 years agoInclude <netinet/in.h> for Solaris 8
Howard Chu [Wed, 15 May 2013 18:50:40 +0000 (11:50 -0700)]
Include <netinet/in.h> for Solaris 8

Needed for <resolv.h>, which is needed for BYTE_ORDER.

11 years agoFix mdb_page_split - nested split
Howard Chu [Tue, 14 May 2013 23:57:50 +0000 (16:57 -0700)]
Fix mdb_page_split - nested split

If updating a page's separator triggers a split in its parent,
we may have missed adjusting the parent's cursor position.

11 years agoDoxygen cleanup
Howard Chu [Mon, 13 May 2013 19:36:47 +0000 (12:36 -0700)]
Doxygen cleanup

11 years agoFix 66c839f0292d9aea12bf93ef6568ac2bd2ca0fc8
Howard Chu [Fri, 10 May 2013 12:56:16 +0000 (05:56 -0700)]
Fix 66c839f0292d9aea12bf93ef6568ac2bd2ca0fc8

Forgot #include <errno.h>

11 years agoRefactor mdb_midl_append, add mdb_midl_grow()
Howard Chu [Fri, 10 May 2013 00:27:29 +0000 (17:27 -0700)]
Refactor mdb_midl_append, add mdb_midl_grow()

11 years agoITS#7515 Fix nested transaction error handling.
Hallvard Furuseth [Sun, 5 May 2013 13:13:31 +0000 (15:13 +0200)]
ITS#7515 Fix nested transaction error handling.

mdb_txn_begin(): Do not free(mt_free_pgs), it needs mdb_midl_free().
mdb_txn_commit(): Catch commit(child) error.

11 years agoFix mdb_env_close(unopened MDB_env).
Hallvard Furuseth [Sun, 5 May 2013 12:26:05 +0000 (14:26 +0200)]
Fix mdb_env_close(unopened MDB_env).

Do not try to scan me_dbxs in a closed/never-opened MDB_env.
Broken by 7d643d3acb94a6d7306330b550fce0e5755444a9
and 151c416b460b02cd96502a47189cda19ab651ada.

11 years agoSimplify/cleanup mdb_cursor_init, C_ALLOCD.
Hallvard Furuseth [Sat, 4 May 2013 10:53:43 +0000 (12:53 +0200)]
Simplify/cleanup mdb_cursor_init, C_ALLOCD.

No real change.
mdb_cursor_init() checks if it needs mx, so pass it unconditionally.
Set C_ALLOCD for shadow cursors, for clarity. (It was always set as
it should anyway from the origin cursor, which would have C_ALLOCD.)

11 years agoMDB warning cleanup.
Hallvard Furuseth [Sat, 4 May 2013 10:41:28 +0000 (12:41 +0200)]
MDB warning cleanup.

Unused function when MDB_DEBUG. Unused 'excl' param.

11 years agomdb_dcmp(): Assume the database has MDB_DUPSORT.
Hallvard Furuseth [Sat, 4 May 2013 10:34:49 +0000 (12:34 +0200)]
mdb_dcmp(): Assume the database has MDB_DUPSORT.

There was little point in returning EINVAL when not: Comparing (A,B)
and (B,A) would claim (A > B && B > A), which could confuse callers.

11 years agoUpdate MDB doc: Cursors, DB handles, data lifetime
Hallvard Furuseth [Sat, 4 May 2013 10:34:20 +0000 (12:34 +0200)]
Update MDB doc: Cursors, DB handles, data lifetime

11 years agoAllow mdb_cursor_close() after readonly txn ends.
Hallvard Furuseth [Sat, 4 May 2013 10:34:02 +0000 (12:34 +0200)]
Allow mdb_cursor_close() after readonly txn ends.

Catch mdb_cursor_renew(write txn's cursor).  Add flag C_UNTRACK, so
mdb_cursor_close need not peek inside a possibly-freed readonly txn.

11 years agoClose cursors when commit(writer). Factor out code.
Hallvard Furuseth [Sat, 4 May 2013 10:33:06 +0000 (12:33 +0200)]
Close cursors when commit(writer). Factor out code.

Close remaining cursors when committing a write txn. The doc says this
happens, and it avoids cursor tracking when updating mainDB + freeDB.

Rename mdb_cursor_merge() -> mdb_cursors_close() for code reuse,
and add a merge option.  Simplify its loop a bit.

Factor out cleanup of DBIs.

11 years agoPlug txn leak and MDB_NOTLS slot leaks on error.
Hallvard Furuseth [Sat, 4 May 2013 10:31:31 +0000 (12:31 +0200)]
Plug txn leak and MDB_NOTLS slot leaks on error.

On mdb_env_copy() error: Abort the txn.
On mdb_txn_renew0() error: Release new MDB_NOTLS reader slot.

11 years agoCleaner "Support mdb_txn_abort(a reset txn)".
Hallvard Furuseth [Sat, 4 May 2013 08:03:19 +0000 (10:03 +0200)]
Cleaner "Support mdb_txn_abort(a reset txn)".

Redo 8a562f560edcc6417672df453c4e626dfb4c4024 so !mt_dbxs indicates
"txn was reset", so mt_numdbs gets one magic value instead of two.

11 years agoUpdate fixes for dbi_open/close, ITS#7515.
Hallvard Furuseth [Sat, 4 May 2013 08:03:19 +0000 (10:03 +0200)]
Update fixes for dbi_open/close, ITS#7515.

Reset me_dbflags[dbi] when closing DBI, to get rid of MDB_VALID flag.
mdb_env_close(): Re-fix DB-name memleak. DBIs > me_numdbs may exist.

11 years agoAllow reading freelist while working on it
Howard Chu [Thu, 2 May 2013 05:43:35 +0000 (22:43 -0700)]
Allow reading freelist while working on it

The circular dependency issues appear to have been resolved.
Still, need to watch closely, maybe revert this change if
problems arise.

11 years agoAvoid assert
Howard Chu [Wed, 1 May 2013 04:09:09 +0000 (21:09 -0700)]
Avoid assert

Due to underfilled branch page. We're in the process of merging/moving
nodes to it because we already know it's underfilled. Took this approach
rather than just removing the assert in mdb_page_search_root, because
that assert may yet catch other situations we don't know about.

(Although, it has been there since the original commit of mdb.c and
has never triggered any other times...)

11 years agoPlug mdb_cursor_renew() memleak
Hallvard Furuseth [Mon, 22 Apr 2013 18:46:56 +0000 (20:46 +0200)]
Plug mdb_cursor_renew() memleak