Кроме небольшого рефакторинга здесь реализуется более регулярный способ
обхода дерева при копировании с компактификаций. В частности, полная
инициализация курсоров позволяет выполнять больше проверок/контроля
структуры БД и избавиться от флажка CC_COPYING.
Beside a small refactoring, a more regular way of traversing the tree
when copying with compactification is implemented here. In particular,
full initialization of cursors allows to perform more checks/control of
the DB structure and get rid of the CC_COPYING flag.
Здесь основная часть изменений преобразующих отладочную проверку страниц
в регулярный и доступный пользователю осторожный/безопасный режим работы
с потенциально поврежденной БД.
Here the major part of the changes that transform a debugging check of
pages into a regular and user-accessible careful/safe mode for working
with a potentially corrupted database.
Chunking long list of retired pages during huge transactions commit
to avoid use sequences of pages:
- splits a long retired page-number-list into chunks
which fits one per single overflow/large page;
- this requires a few unique id for keys
for create such records into GC/freeDB;
- just use the necessary subsequent IDs following the current
transaction ID and then take the last of ones to update a meta-page.
Thus avoids using/allocating/searching a sequence of free pages
but just increase txnid more than one during the commit
a huge write transaction with a long retired-pages-list.
The stable release with an important fixes and workaround for the critical macOS thread-local-storage issue.
Acknowledgements:
-----------------
- [Masatoshi Fukunaga](https://github.com/mah0x211) for [Lua bindings](https://github.com/mah0x211/lua-libmdbx).
New:
----
- Added most of transactions flags to the public API.
- Added `MDBX_NOSUCCESS_EMPTY_COMMIT` build option to return non-success result (`MDBX_RESULT_TRUE`) on empty commit.
- Reworked validation and import of DBI-handles into a transaction.
Assumes these changes will be invisible to most users, but will cause fewer surprises in complex DBI cases.
- Added ability to open DB in without-LCK (exclusive read-only) mode in case no permissions to create/write LCK-file.
Fixes:
------
- A series of fixes and improvements for automatically generated documentation (Doxygen).
- Fixed copy&paste bug with could lead to `SIGSEGV` (nullptr dereference) in the exclusive/no-lck mode.
- Fixed minor warnings from modern Apple's CLANG 13.
- Fixed minor warnings from CLANG 14 and in-development CLANG 15.
- Fixed `SIGSEGV` regression in without-LCK (exclusive read-only) mode.
- Fixed `mdbx_check_fs_local()` for CDROM case on Windows.
- Fixed nasty typo of typename which caused false `MDBX_CORRUPTED` error in a rare execution path,
when the size of the thread-ID type not equal to 8.
- Fixed write-after-free memory corruption on latest `macOS` during finalization/cleanup of thread(s) that executed read transaction(s).
> The issue was suddenly discovered by a [CI](https://en.wikipedia.org/wiki/Continuous_integration)
> after adding an iteration with macOS 11 "Big Sur", and then reproduced on recent release of macOS 12 "Monterey".
> The issue was never noticed nor reported on macOS 10 "Catalina" nor others.
> Analysis shown that the problem caused by a change in the behavior of the system library (internals of dyld and pthread)
> during thread finalization/cleanup: now a memory allocated for a `__thread` variable(s) is released
> before execution of the registered Thread-Local-Storage destructor(s),
> thus a TLS-destructor will write-after-free just by legitime dereference any `__thread` variable.
> This is unexpected crazy-like behavior since the order of resources releasing/destroying
> is not the reverse of ones acquiring/construction order. Nonetheless such surprise
> is now workarounded by using atomic compare-and-swap operations on a 64-bit signatures/cookies.
- Fixed Elbrus/E2K LCC 1.26 compiler warnings (memory model for atomic operations, etc).
Minors:
-------
- Refined `release-assets` GNU Make target.
- Added logging to `mdbx_fetch_sdb()` to help debugging complex DBI-handels use cases.
- Added explicit error message from probe of no-support for `std::filesystem`.
- Added contributors "score" table by `git fame` to generated docs.
- Added `mdbx_assert_fail()` to public API (mostly for backtracing).
- Now C++20 concepts used/enabled only when `__cpp_lib_concepts >= 202002`.
- Don't provide nor report package information if used as a CMake subproject.
Signed-off-by: Леонид Юрьев (Leonid Yuriev) <leo@yuriev.ru>
The stable risen release after the Github's intentional malicious disaster.
We have migrated to a reliable trusted infrastructure
-----------------------------------------------------
The origin for now is at [GitFlic](https://gitflic.ru/project/erthink/libmdbx)
since on 2022-04-15 the Github administration, without any warning nor
explanation, deleted _libmdbx_ along with a lot of other projects,
simultaneously blocking access for many developers.
For the same reason ~~Github~~ is blacklisted forever.
GitFlic already support Russian and English languages, plan to support more,
including 和 中文. You are welcome!
New:
----
- Added the `tools-static` make target to build statically linked MDBX tools.
- Support for Microsoft Visual Studio 2022.
- Support build by MinGW' make from command line without CMake.
- Added `mdbx::filesystem` C++ API namespace that corresponds to `std::filesystem` or `std::experimental::filesystem`.
- Created [website](https://libmdbx.website.yandexcloud.net/) for online auto-generated documentation.
- Used `todo4recovery://erased_by_github/` for dead (or temporarily lost) resources deleted by ~~Github~~.
- Added `--loglevel=` command-line option to the `mdbx_test` tool.
- Added few fast smoke-like tests into CMake builds.
Fixes:
------
- Fixed a race between starting a transaction and creating a DBI descriptor that could lead to `SIGSEGV` in the cursor tracking code.
- Clarified description of `MDBX_EPERM` error returned from `mdbx_env_set_geometry()`.
- Fixed non-promoting the parent transaction to be dirty in case the undo of the geometry update failed during abortion of a nested transaction.
- Resolved linking issues with `libstdc++fs`/`libc++fs`/`libc++experimental` for C++ `std::filesystem` or `std::experimental::filesystem` for legacy compilers.
- Added workaround for GNU Make 3.81 and earlier.
- Added workaround for Elbrus/LCC 1.25 compiler bug of class inline `static constexpr` member field.
- [Fixed](https://github.com/ledgerwatch/erigon/issues/3874) minor assertion regression (only debug builds were affected).
- Fixed detection of `C++20` concepts accessibility.
- Fixed detection of Clang's LTO availability for Android.
- Fixed build for ARM/ARM64 by MSVC.
- Fixed non-x86 Windows builds with `MDBX_WITHOUT_MSVC_CRT=ON` and `MDBX_BUILD_SHARED_LIBRARY=ON`.
Minors:
-------
- Resolve minor MSVC warnings: avoid `/INCREMENTAL[:YES]` with `/LTCG`, `/W4` with `/W3`, the `C5105` warning.
- Switched to using `MDBX_EPERM` instead of `MDBX_RESULT_TRUE' to indicate that the geometry cannot be updated.
- Added `NULL` checking during memory allocation inside `mdbx_chk`.
- Resolved all warnings from MinGW while used without CMake.
- Added inheretable `target_include_directories()` to `CMakeLists.txt` for easy integration.
- Added build-time checks and paranoid runtime assertions for the `off_t` arguments of `fcntl()` which are used for locking.
- Added `-Wno-lto-type-mismatch` to avoid false-positive warnings from old GCC during LTO-enabled builds.
- Added checking for TID (system thread id) to avoid hang on 32-bit Bionic/Android within `pthread_mutex_lock()`.
- Reworked `MDBX_BUILD_TARGET` of CMake builds.
- Added `CMAKE_HOST_ARCH` and `CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET`.
Signed-off-by: Леонид Юрьев (Leonid Yuriev) <leo@yuriev.ru>
Fixes https://github.com/ledgerwatch/erigon/issues/3874.
This was a minor regression after the c4a5325aafd3f03ce7520731b9da7253d7d178f0
that affects only debug builgs (with enabled assertions) and only when the added
code catch a incoherency of unified page/buffer cache.
The error was that the array of pointers in the transaction zeroed by the
value of env->me_numdbs and before txn->mt_numdbs was set to env->me_numdbs.
Thus, a cursor pointer(s) in the starting transaction could uninitialized if
another thread opened a dbi-handle(s) between the two aforementioned events.
The stable release with the complete workaround for an incoherence flaw of Linux unified page/buffer cache.
Nonetheless the cause for this trouble may be an issue of Intel CPU cache/MESI.
See [issue#269](https://github.com/erthink/libmdbx/issues/269) for more information.
Acknowledgements:
-----------------
- [David Bouyssié](https://github.com/david-bouyssie) for [Scala bindings](https://github.com/david-bouyssie/mdbx4s).
- [Michelangelo Riccobene](https://github.com/mriccobene) for reporting and testing.
Fixes:
------
- [Added complete workaround](https://github.com/erthink/libmdbx/issues/269) for an incoherence flaw of Linux unified page/buffer cache.
- [Fixed](https://github.com/erthink/libmdbx/issues/272) cursor reusing for read-only transactions.
- Fixed copy&paste typo inside `mdbx::cursor::find_multivalue()`.
Minors:
-------
- Minor refine C++ API for convenience.
- Minor internals refines.
- Added `lib-static` and `lib-shared` targets for make.
- Added minor workaround for AppleClang 13.3 bug.
- Clarified error messages of a signature/version mismatch.
Signed-off-by: Леонид Юрьев (Leonid Yuriev) <leo@yuriev.ru>
Briefly, this commit fixes a missed flaw:
- Cursor tracking is required to replacing shaded pages and adjusting the positions in writing transactions;
- Thus, historically, an internal linked list was maintained for a read-write transactions, but not for a read-only.
For this reason, the API for using cursors should be different for writing and reading transactions;
- However, the libmdbx's API has been significantly improved, including the ability to reuse cursors and a uniform cursors behavior for any kind of transactions.
My mistake is that due to working with MithrilDB, I forgot to make a same changes to libmdbx.
Fixes https://github.com/erthink/libmdbx/issues/272.
The stable release with the hotfix/workaround for a flaw of Linux 4.19 (at least) unified page/buffer cache.
See [issue#269](https://github.com/erthink/libmdbx/issues/269) for more information.
Acknowledgements:
-----------------
- [Simon Leier](https://github.com/leisim) for reporting and testing.
- [Kai Wetlesen](https://github.com/kaiwetlesen) for [RPMs](http://copr.fedorainfracloud.org/coprs/kwetlesen/libmdbx/).
- [Tullio Canepa](https://github.com/canepat) for reporting C++ API issue and contributing.
Fixes:
------
- [Added workaround](https://github.com/erthink/libmdbx/issues/269) for a flaw of Linux 4.19 (at least) unified page/buffer cache.
- [Fixed/Reworked](https://github.com/erthink/libmdbx/pull/270) move-assignment operators for "managed" classes of C++ API.
- Fixed potential `SIGSEGV` while open DB with overrided non-default page size.
- [Made](https://github.com/erthink/libmdbx/issues/267) `mdbx_env_open()` idempotence in failure cases.
- Refined/Fixed pages reservation inside `mdbx_update_gc()` to avoid non-reclamation in a rare cases.
- Fixed typo in a retained space calculation for the hsr-callback.
Minors:
-------
- Reworked functions for meta-pages, split-off non-volatile.
- Disentangled C11-atomic fences/barriers and pure-functions (with `__attribute__((__pure__))`) to avoid compiler misoptimization.
- Fixed hypotetic unaligned access to 64-bit dwords on ARM with `__ARM_FEATURE_UNALIGNED` defined.
- Reasonable paranoia that makes clarity for code readers.
- Minor fixes Doxygen references, comments, descriptions, etc.
Signed-off-by: Леонид Юрьев (Leonid Yuriev) <leo@yuriev.ru>
The three points:
- disentangle C11-atomic fences/barriers and pure-functions (with `__attribute__((__pure__))`) to avoid compiler misoptimization;
- fix hypotetic unaligned access to 64-bit dwords on ARM with `__ARM_FEATURE_UNALIGNED` defined;
- reasonable paranoia that makes clarity for code readers.
The stable release with fixes for large and huge databases sized of 4..128 TiB.
Acknowledgements:
-----------------
- Ledgerwatch, Binance and Positive Technologies teams for reporting, assistance in investigation and testing.
- Alex Sharov for reporting, testing and provide resources for remote debugging/investigation.
- Kris Zyp for Deno support.
New features, extensions and improvements:
------------------------------------------
- Added treating the `UINT64_MAX` value as maximum for given option inside `mdbx_env_set_option()`.
- Added `to_hex/to_base58/to_base64::output(std::ostream&)` overloads without using temporary string objects as buffers.
- Added `--geometry-jitter=YES|no` option to the test framework.
- Added support for [Deno](https://deno.land/) support by [Kris Zyp](https://github.com/kriszyp).
Fixes:
------
- Fixed handling `MDBX_opt_rp_augment_limit` for GC's records from huge transactions (Erigon/Akula/Ethereum).
- [Fixed](https://github.com/erthink/libmdbx/issues/258) build on Android (avoid including `sys/sem.h`).
- [Fixed](https://github.com/erthink/libmdbx/pull/261) missing copy assignment operator for `mdbx::move_result`.
- Fixed missing `&` for `std::ostream &operator<<()` overloads.
- Fixed unexpected `EXDEV` (Cross-device link) error from `mdbx_env_copy()`.
- Fixed base64 encoding/decoding bugs in auxillary C++ API.
- Fixed overflow of `pgno_t` during checking PNL on 64-bit platforms.
- [Fixed](https://github.com/erthink/libmdbx/issues/260) excessive PNL checking after sort for spilling.
- Reworked checking `MAX_PAGENO` and DB upper-size geometry limit.
- [Fixed](https://github.com/erthink/libmdbx/issues/265) build for some combinations of versions of MSVC and Windows SDK.
Minors:
-------
- Added workaround for CLANG bug [D79919/PR42445](https://reviews.llvm.org/D79919).
- Fixed build test on Android (using `pthread_barrier_t` stub).
- Disabled C++20 concepts for CLANG < 14 on Android.
- Fixed minor `unused parameter` warning.
- Added CI for Android.
- Refine/cleanup internal logging.
- Refined line splitting inside hex/base58/base64 encoding to avoid `\n` at the end.
- Added workaround for modern libstdc++ with CLANG < 4.x
- Relaxed txn-check rules for auxiliary functions.
- Clarified a comments and descriptions, etc.
- Using the `-fno-semantic interposition` option to reduce the overhead to calling self own public functions.
Signed-off-by: Леонид Юрьев (Leonid Yuriev) <leo@yuriev.ru>
This bug triggered only in the DEBUG builds or when the assertion checking is forcibly enabled.
It does not affect any core logic and cannot lead to DB corruption, data loss, and so on.
Fixes https://github.com/erthink/libmdbx/issues/260.
Added a check that the data of the BIGDATA node (containing the target page number) is located within the boundaries of the page being checked.
The third case of https://github.com/erthink/libmdbx/issues/217.
Here are some changes to avoid recursive acquisition of SRW-lock,
which is still in use:
- Read transactions don't acquire the shared SRW-lock with `MDBX_NOTLS.
- Memory-mapping of DB is always kept while DB opened,
therefore following limitations are:
- DB file can't be shrinked while it used,
including auto-shrink due to auto-compactification with corresponding geometry settings.
- The upper limit of DB size can't be changed while DB is used.
- The DB can grow within the upper size limit defined while opening by a first process,
but this does not work under Wine since there is no `NtExtendSection()` function.
Partially fix https://github.com/erthink/libmdbx/issues/203
Acknowledgements:
-----------------
- [Mahlon E. Smith](https://github.com/mahlonsmith) for [Ruby bindings](https://rubygems.org/gems/mdbx/).
- [Alex Sharov](https://github.com/AskAlexSharov) for [mdbx-go](https://github.com/torquem-ch/mdbx-go), bug reporting and testing.
- [Artem Vorotnikov](https://github.com/vorot93) for bug reporting and PR.
- [Paolo Rebuffo](https://www.linkedin.com/in/paolo-rebuffo-8255766/), [Alexey Akhunov](https://github.com/AlexeyAkhunov) and Mark Grosberg for donations.
- [Noel Kuntze](https://github.com/Thermi) for preliminary [Python bindings](https://github.com/Thermi/libmdbx/tree/python-bindings)
New features:
-------------
- Added `mdbx_env_set_option()` and `mdbx_env_get_option()` for controls
various runtime options for an environment (announce of this feature was missed in a previous news).
- Added `MDBX_DISABLE_PAGECHECKS` build option to disable some checks to reduce an overhead
and detection probability of database corruption to a values closer to the LMDB.
The `MDBX_DISABLE_PAGECHECKS=1` provides a performance boost of about 10% in CRUD scenarios,
and conjointly with the `MDBX_ENV_CHECKPID=0` and `MDBX_TXN_CHECKOWNER=0` options can yield
up to 30% more performance compared to LMDB.
- Using float point (exponential quantized) representation for internal 16-bit values
of grow step and shrink threshold when huge ones (https://github.com/erthink/libmdbx/issues/166).
To minimize the impact on compatibility, only the odd values inside the upper half
of the range (i.e. 32769..65533) are used for the new representation.
- Added the `mdbx_drop` similar to LMDB command-line tool to purge or delete (sub)database(s).
- [Ruby bindings](https://rubygems.org/gems/mdbx/) is available now by [Mahlon E. Smith](https://github.com/mahlonsmith).
- Added `MDBX_ENABLE_MADVISE` build option which controls the use of POSIX `madvise()` hints and friends.
- The internal node sizes were refined, resulting in a reduction in large/overflow pages in some use cases
and a slight increase in limits for a keys size to ≈½ of page size.
- Added to `mdbx_chk` output number of keys/items on pages.
- Added explicit `install-strip` and `install-no-strip` targets to the `Makefile` (https://github.com/erthink/libmdbx/pull/180).
- Major rework page splitting (af9b7b560505684249b76730997f9e00614b8113) for
- An "auto-appending" feature upon insertion for both ascending and
descending key sequences. As a result, the optimality of page filling
increases significantly (more densely, less slackness) while
inserting ordered sequences of keys,
- A "splitting at middle" to make page tree more balanced on average.
- Added `mdbx_get_sysraminfo()` to the API.
- Added guessing a reasonable maximum DB size for the default upper limit of geometry (https://github.com/erthink/libmdbx/issues/183).
- Major rework internal labeling of a dirty pages (958fd5b9479f52f2124ab7e83c6b18b04b0e7dda) for
a "transparent spilling" feature with the gist to make a dirty pages
be ready to spilling (writing to a disk) without further altering ones.
Thus in the `MDBX_WRITEMAP` mode the OS kernel able to oust dirty pages
to DB file without further penalty during transaction commit.
As a result, page swapping and I/O could be significantly reduced during extra large transactions and/or lack of memory.
- Minimized reading leaf-pages during dropping subDB(s) and nested trees.
- Major rework a spilling of dirty pages to support [LRU](https://en.wikipedia.org/wiki/Cache_replacement_policies#Least_recently_used_(LRU))
policy and prioritization for a large/overflow pages.
- Statistics of page operations (split, merge, copy, spill, etc) now available through `mdbx_env_info_ex()`.
- Auto-setup limit for length of dirty pages list (`MDBX_opt_txn_dp_limit` option).
- Support `make options` to list available build options.
- Support `make help` to list available make targets.
- Silently `make`'s build by default.
- Preliminary [Python bindings](https://github.com/Thermi/libmdbx/tree/python-bindings) is available now
by [Noel Kuntze](https://github.com/Thermi) (https://github.com/erthink/libmdbx/issues/147).
Backward compatibility break:
-----------------------------
- The `MDBX_AVOID_CRT` build option was renamed to `MDBX_WITHOUT_MSVC_CRT`.
This option is only relevant when building for Windows.
- The `mdbx_env_stat()` always, and `mdbx_env_stat_ex()` when called with the zeroed transaction parameter,
now internally start temporary read transaction and thus may returns `MDBX_BAD_RSLOT` error.
So, just never use deprecated `mdbx_env_stat()' and call `mdbx_env_stat_ex()` with transaction parameter.
- The build option `MDBX_CONFIG_MANUAL_TLS_CALLBACK` was removed and now just a non-zero value of
the `MDBX_MANUAL_MODULE_HANDLER` macro indicates the requirement to manually call `mdbx_module_handler()`
when loading libraries and applications uses statically linked libmdbx on an obsolete Windows versions.
Fixes:
------
- Fixed performance regression due non-optimal C11 atomics usage (https://github.com/erthink/libmdbx/issues/160).
- Fixed "reincarnation" of subDB after it deletion (https://github.com/erthink/libmdbx/issues/168).
- Fixed (disallowing) implicit subDB deletion via operations on `@MAIN`'s DBI-handle.
- Fixed a crash of `mdbx_env_info_ex()` in case of a call for a non-open environment (https://github.com/erthink/libmdbx/issues/171).
- Fixed the selecting/adjustment values inside `mdbx_env_set_geometry()` for implicit out-of-range cases (https://github.com/erthink/libmdbx/issues/170).
- Fixed `mdbx_env_set_option()` for set initial and limit size of dirty page list ((https://github.com/erthink/libmdbx/issues/179).
- Fixed an unreasonably huge default upper limit for DB geometry (https://github.com/erthink/libmdbx/issues/183).
- Fixed `constexpr` specifier for the `slice::invalid()`.
- Fixed (no)readahead auto-handling (https://github.com/erthink/libmdbx/issues/164).
- Fixed non-alloy build for Windows.
- Switched to using Heap-functions instead of LocalAlloc/LocalFree on Windows.
- Fixed `mdbx_env_stat_ex()` to returning statistics of the whole environment instead of MainDB only (https://github.com/erthink/libmdbx/issues/190).
- Fixed building by GCC 4.8.5 (added workaround for a preprocessor's bug).
- Fixed building C++ part for iOS <= 13.0 (unavailability of `std::filesystem::path`).
- Fixed building for Windows target versions prior to Windows Vista (`WIN32_WINNT < 0x0600`).
- Fixed building by MinGW for Windows (https://github.com/erthink/libmdbx/issues/155).
TODO for a next releases:
-------------------------
- [Get rid of dirty-pages list in MDBX_WRITEMAP mode](https://github.com/erthink/libmdbx/issues/193).
- [Large/Overflow pages accounting for dirty-room](https://github.com/erthink/libmdbx/issues/192).
- [C++ Buffer issue](https://github.com/erthink/libmdbx/issues/191).
- Finalize C++ API (few typos and trivia bugs are still likely for now).
- [Support for RAW devices](https://github.com/erthink/libmdbx/issues/124).
- [Test framework issue](https://github.com/erthink/libmdbx/issues/127).
- [Support MessagePack for Keys & Values](https://github.com/erthink/libmdbx/issues/115).
- [Engage new terminology](https://github.com/erthink/libmdbx/issues/137).
- Packages for [Astra Linux](https://astralinux.ru/), [ALT Linux](https://www.altlinux.org/), [ROSA Linux](https://www.rosalinux.ru/), Fedora/RHEL, Debian/Ubuntu.
Briefly:
- Now constructor/destructor of "Thread Local Storage" handled automatically when possible.
- Otherwise the MDBX_CONFIG_MANUAL_TLS_CALLBACK macro defined to 1 to indicate that mdbx_module_handle() should be called manually.
- Corresponding build option MDBX_CONFIG_MANUAL_TLS_CALLBACK was removed.
Related to https://github.com/erthink/libmdbx/issues/155
Change-Id: Ic4e6a34b44f874676f0ab212ff473460e3d80559
Resolves https://github.com/erthink/libmdbx/issues/164
---
NOTE: Seems there is a bug in the Mach/Darwin/OSX kernel,
because MADV_WILLNEED with offset != 0 may cause SIGBUS
on following access to the hinted region.
19.6.0 Darwin Kernel Version 19.6.0: Tue Jan 12 22:13:05 PST 2021; root:xnu-6153.141.16~1/RELEASE_X86_64 x86_64
Change-Id: I11ebbf2bd35e3dba9d078be16cb5678aecf8329c
Basically, this (squashed) commit introduces:
- An "auto-appending" feature upon insertion for both ascending and
descending key sequences. As a result, the optimality of page filling
increases significantly (more densely, less slackness) while
inserting ordered sequences of keys,
- A "splitting at middle" for more balanced page tree on average.
---
1. Using left/middle/right tactics for finding the split point of a page:
- If a key is inserted close to an edge of page,
then the page splits at that edge;
- Otherwise a page splits at the middle,
which leads to a more balanced tree on average;
- So I expect a better behavior on average,
but actually effects should be studied further practically.
2. New code for calculating the midpoint of a page split.
3. APPEND-flags no longer affect choosing the page split point.
4. Added left-side splitting by inserting a pure page with a new entry.
Change-Id: Id7441acfc8c90636e3be6bc00a0df15714690f3c
Using float point (exponential quantized) representation for internal 16-bit values
of grow step and shrink threshold when huge ones
.
To minimize the impact on compatibility, only the odd values inside the upper half
of the range (i.e. 32769..65533) are used for the new representation.
Resolve https://github.com/erthink/libmdbx/issues/166
Change-Id: I273127c1842deef0d7d8885b55a805b1463556eb
Acknowledgements:
-----------------
- [Mahlon E. Smith](http://www.martini.nu/) for [FreeBSD port of libmdbx](https://svnweb.freebsd.org/ports/head/databases/mdbx/).
- [장세연](http://www.castis.com) for bug fixing and PR.
- [Clément Renault](https://github.com/Kerollmops/heed) for [Heed](https://github.com/Kerollmops/heed) fully typed Rust wrapper.
- [Alex Sharov](https://github.com/AskAlexSharov) for bug reporting.
- [Noel Kuntze](https://github.com/Thermi) for bug reporting.
Removed options and features:
-----------------------------
- Drop `MDBX_HUGE_TRANSACTIONS` build-option (now no longer required).
New features:
-------------
- Package for FreeBSD is available now by Mahlon E. Smith.
- New API functions to get/set various options (https://github.com/erthink/libmdbx/issues/128):
- the maximum number of named databases for the environment;
- the maximum number of threads/reader slots;
- threshold (since the last unsteady commit) to force flush the data buffers to disk;
- relative period (since the last unsteady commit) to force flush the data buffers to disk;
- limit to grow a list of reclaimed/recycled page's numbers for finding a sequence of contiguous pages for large data items;
- limit to grow a cache of dirty pages for reuse in the current transaction;
- limit of a pre-allocated memory items for dirty pages;
- limit of dirty pages for a write transaction;
- initial allocation size for dirty pages list of a write transaction;
- maximal part of the dirty pages may be spilled when necessary;
- minimal part of the dirty pages should be spilled when necessary;
- how much of the parent transaction dirty pages will be spilled while start each child transaction;
- Unlimited/Dynamic size of retired and dirty page lists (https://github.com/erthink/libmdbx/issues/123).
- Added `-p` option (purge subDB before loading) to `mdbx_load` tool.
- Reworked spilling of large transaction and committing of nested transactions:
- page spilling code reworked to avoid the flaws and bugs inherited from LMDB;
- limit for number of dirty pages now is controllable at runtime;
- a spilled pages, including overflow/large pages, now can be reused and refunded/compactified in nested transactions;
- more effective refunding/compactification especially for the loosed page cache.
- Added `MDBX_ENABLE_REFUND` and `MDBX_PNL_ASCENDING` internal/advanced build options.
- Added `mdbx_default_pagesize()` function.
- Better support architectures with a weak/relaxed memory consistency model (ARM, AARCH64, PPC, MIPS, RISC-V, etc) by means [C11 atomics](https://en.cppreference.com/w/c/atomic).
- Speed up page number lists and dirty page lists (https://github.com/erthink/libmdbx/issues/132).
- Added `LIBMDBX_NO_EXPORTS_LEGACY_API` build option.
Fixes:
------
- Fixed missing cleanup (null assigned) in the C++ commit/abort (https://github.com/erthink/libmdbx/pull/143).
- Fixed `mdbx_realloc()` for case of nullptr and `MDBX_AVOID_CRT=ON` for Windows.
- Fixed the possibility to use invalid and renewed (closed & re-opened, dropped & re-created) DBI-handles (https://github.com/erthink/libmdbx/issues/146).
- Fixed 4-byte aligned access to 64-bit integers, including access to the `bootid` meta-page's field (https://github.com/erthink/libmdbx/issues/153).
- Fixed minor/potential memory leak during page flushing and unspilling.
- Fixed handling states of cursors's and subDBs's for nested transactions.
- Fixed page leak in extra rare case the list of retired pages changed during update GC on transaction commit.
- Fixed assertions to avoid false-positive UB detection by CLANG/LLVM (https://github.com/erthink/libmdbx/issues/153).
- Fixed `MDBX_TXN_FULL` and regressive `MDBX_KEYEXIST` during large transaction commit with `MDBX_LIFORECLAIM` (https://github.com/erthink/libmdbx/issues/123).
- Fixed auto-recovery (`weak->steady` with the same boot-id) when Database size at last weak checkpoint is large than at last steady checkpoint.
- Fixed operation on systems with unusual small/large page size, including PowerPC (https://github.com/erthink/libmdbx/issues/157).
TODO:
-----
- Engage new terminology (https://github.com/erthink/libmdbx/issues/137).
- Resolve few TODOs (https://github.com/erthink/libmdbx/issues/124, https://github.com/erthink/libmdbx/issues/127, https://github.com/erthink/libmdbx/issues/115).
- Finalize C++ API.
- Packages for [ROSA Linux](https://www.rosalinux.ru/), [ALT Linux](https://www.altlinux.org/), Fedora/RHEL, Debian/Ubuntu.
Change-Id: I414b8ef2e4b90e04fb344779c0e3f1b4bd1c06be
This done better support architectures with a weak/relaxed memory consistency model (ARM, AARCH64, PPC, MIPS, RISC-V, etc).
Change-Id: Iee831c8dc564f1d027ff84b0d6daa559325d5a9b
Fix regression related to https://github.com/erthink/libmdbx/issues/123 and https://github.com/erthink/libmdbx/issues/128.
Related to https://github.com/erthink/libmdbx/issues/131.
В lifo-режиме при фиксации транзакции, записи в GC могли быть перезаписаны (с утечкой страниц БД), либо могла возникать ошибка MDBX_KEYEXISTS, по следующему сценарию:
- В истории БД были две транзакции с огромным кол-вом retired pages, после которых в GC остались две соответствующие записи.
- В ходе очередной транзакции первая из огромных GC-записей попадает в переработку и образует огромный reclaimed list.
- При фиксации транзакции производится попытка разбить огромный reclaimed list на чанки размером в одну страницу. Для этого требуется много id для записей, которые в соответствии с lifo должны быть максимально близки к голове GC, т. е. получены путем переработки последних записей GC.
- В ходе переработки последних записей очередь доходит до второй огромной записи, при этом переработка прерывается, ибо иначе reclaimed list переполнится.
- Однако прерывание переработки внутри mdbx_update_gc() трактовалось как отсутствие записей в GC, поэтому список доступных просто добавлялись соответствующие id-шники.
- Если в списке доступных id-шников для помещения в GC были переработанные, то записи с id по всему списку удалялись - тогда вторая большая запись (и возможно предыдущие) удалялись, а содержащиеся в них номера страниц выпадали из оборота.
- Если же в списке доступных id-шников не было переработанных, то чистка не проводилась - тогда при последующая попытка помещения чанков reclaimed list в GC завершалась ошибкой MDBX_KEYEXISTS, которая и возвращалась из mdbx_commit_ex().
Change-Id: I3e5d40ef7950b7476da0513c6836fcba1de74879
Historically, the page header provides 4-byte data alignment.
Therefore, unfortunately, the meta page data is also aligned on a 4-byte boundary, but contains 64-bit values.
This commit eliminates potentially unsafe access (SPARC, MIPS, etc) to these 64-bit values aligned on a 4-byte boundary.
Thus, a build with the `-fsanitize=undefined` now passes the tests both with CLANG 11 and GCC 10.
Change-Id: Ie441103e53ed96fd40507d8c0be0689e3fee69f5
