Summary:
These are optimization that we applied to improve sysbech's update_noindex performance.
1. Make use of LIKELY compiler hint
2. Move std::atomic so the subclass
3. Make use of skip_prepared in non-2pc transactions.
Closes https://github.com/facebook/rocksdb/pull/3512

Differential Revision: D7000075

Pulled By: maysamyabandeh

fbshipit-source-id: 1ab8292584df1f6305a4992973fb1b7933632181

Summary:
Deadlock: a memtable flush holds DB::mutex_ and calls ThreadLocalPtr::Scrape(), which locks ThreadLocalPtr mutex; meanwhile, a thread exit handler locks ThreadLocalPtr mutex and calls SuperVersionUnrefHandle, which tries to lock DB::mutex_.

This deadlock is hit all the time on our workload. It blocks our release.

In general, the problem is that ThreadLocalPtr takes an arbitrary callback and calls it while holding a lock on a global mutex. The same global mutex is (at least in some cases) locked by almost all ThreadLocalPtr methods, on any instance of ThreadLocalPtr. So, there'll be a deadlock if the callback tries to do anything to any instance of ThreadLocalPtr, or waits for another thread to do so.

So, probably the only safe way to use ThreadLocalPtr callbacks is to do only do simple and lock-free things in them.

This PR fixes the deadlock by making sure that local_sv_ never holds the last reference to a SuperVersion, and therefore SuperVersionUnrefHandle never has to do any nontrivial cleanup.

I also searched for other uses of ThreadLocalPtr to see if they may have similar bugs. There's only one other use, in transaction_lock_mgr.cc, and it looks fine.
Closes https://github.com/facebook/rocksdb/pull/3510

Reviewed By: sagar0

Differential Revision: D7005346

Pulled By: al13n321

fbshipit-source-id: 37575591b84f07a891d6659e87e784660fde815f

Summary:
The MemTableRep API was broken by this commit: 813719e9
This patch reverts the changes and instead adds InsertKey (and etc.) overloads to extend the MemTableRep API without breaking the existing classes that inherit from it.
Closes https://github.com/facebook/rocksdb/pull/3513

Differential Revision: D7004134

Pulled By: maysamyabandeh

fbshipit-source-id: e568d91fe1e17dd76c0c1f6c7dd51a18633b1c4f

Summary:
- removed a few unneeded variables
- fused some variable declarations and their assignments
- fixed right-trimming code in string_util.cc to not underflow
- simplifed an assertion
- move non-nullptr check assertion before dereferencing of that pointer
- pass an std::string function parameter by const reference instead of by value (avoiding potential copy)
Closes https://github.com/facebook/rocksdb/pull/3507

Differential Revision: D7004679

Pulled By: sagar0

fbshipit-source-id: 52944952d9b56dfcac3bea3cd7878e315bb563c4

Summary:
Right now it is possible that a file gets assigned to L0 but also assigned the seqno from a higher level which it doesn't fit
Under the current impl, it is possibe that seqno in lower levels (Ln) can be equal to smallest seqno of higher levels (Ln-1), which is undesirable from universal compaction's point of view.
This should fix the intermittent failure of `ExternalSSTFileBasicTest.IngestFileWithGlobalSeqnoPickedSeqno`
Closes https://github.com/facebook/rocksdb/pull/3411

Differential Revision: D6813802

Pulled By: miasantreble

fbshipit-source-id: 693d0462fa94725ccfb9d8858743e6d2d9992d14

Summary:
As in #3425
Closes https://github.com/facebook/rocksdb/pull/3497

Differential Revision: D6979588

Pulled By: gfosco

fbshipit-source-id: e9fb32d04ad45575dfe9de1d79348d158e474197

Summary:
This fixes shift and signed-integer-overflow UBSAN checks in fault_injection_test by using a larger and unsigned type.
Closes https://github.com/facebook/rocksdb/pull/3498

Reviewed By: siying

Differential Revision: D6981116

Pulled By: igorsugak

fbshipit-source-id: 3688f62cce570534b161e9b5f42109ebc9ae5a2c

Summary:
A new LevelIterator was recently created. Rename the old one to make unity build happy. It's also not a good idea to have two classes in the same name anyway.
Closes https://github.com/facebook/rocksdb/pull/3499

Differential Revision: D6979325

Pulled By: siying

fbshipit-source-id: 3a032d93fe205650a08e92e5262594731ec726bb

Summary:
Use a customzied BlockBasedTableIterator and LevelIterator to replace current implementations leveraging two-level-iterator. Hope the customized logic will make code easier to understand. As a side effect, BlockBasedTableIterator reduces the allocation for the data block iterator object, and avoid the virtual function call to it, because we can directly reference BlockIter, a final class. Similarly, LevelIterator reduces virtual function call to the dummy iterator iterating the file metadata. It also enabled further optimization.

The upper bound check is also moved from index block to data block. This implementation fits this iterator better. After the change, forwared iterator is slightly optimized to ensure we trim those iterators.

The two-level-iterator now is only used by partitioned index, so it is simplified.
Closes https://github.com/facebook/rocksdb/pull/3406

Differential Revision: D6809041

Pulled By: siying

fbshipit-source-id: 7da3b9b1d3c8e9d9405302c15920af1fcaf50ffa

Summary:
- Refactored logic for checking write stall condition to a helper function: `GetWriteStallConditionAndCause`. Now it is decoupled from the logic for updating WriteController / stats in `RecalculateWriteStallConditions`, so we can reuse it for predicting whether write stall will occur.
- Updated `CompactRange` to first check whether the one additional immutable memtable / L0 file would cause stalling before it flushes. If so, it waits until that is no longer true.
- Updated `bg_cv_` to be signaled on `SetOptions` calls. The stall conditions `CompactRange` cares about can change when (1) flush finishes, (2) compaction finishes, or (3) options dynamically change. The cv was already signaled for (1) and (2) but not yet for (3).
Closes https://github.com/facebook/rocksdb/pull/3381

Differential Revision: D6754983

Pulled By: ajkr

fbshipit-source-id: 5613e03f1524df7192dc6ae885d40fd8f091d972

Summary: Closes https://github.com/facebook/rocksdb/pull/3487

Differential Revision: D6967098

Pulled By: miasantreble

fbshipit-source-id: 48e0accf2e3b3f589ddb797ff8083c8520269bf0

Summary:
Right now, users will encounter unexpected bahavior if they use key or value larger than 4GB. We should explicitly fail the queriers.
Closes https://github.com/facebook/rocksdb/pull/3484

Differential Revision: D6953895

Pulled By: siying

fbshipit-source-id: b60491e1af064fc5d52971956661f6c18ceac24f

Summary:
CompactionIterator invoke MergeHelper::MergeUntil() to do partial merge between snapshot boundaries. Previously it only depend on sequence number to tell snapshot boundary, but we also need to make use of snapshot_checker to verify visibility of the merge operands to the snapshots. For example, say there is a snapshot with seq = 2 but only can see data with seq <= 1. There are three merges, each with seq = 1, 2, 3. A correct compaction output would be (1),(2+3). Without taking snapshot_checker into account when generating merge result, compaction will generate output (1+2),(3).

By filtering uncommitted keys with read callback, the read path already take care of merges well and don't need additional updates.
Closes https://github.com/facebook/rocksdb/pull/3475

Differential Revision: D6926087

Pulled By: yiwu-arbug

fbshipit-source-id: 8f539d6f897cfe29b6dc27a8992f68c2a629d40a

Summary:
It's always a mystery from the logs why flush was triggered -- user triggered it manually, WriteBufferManager triggered it,  logs were full, write buffer was full, etc.
This PR logs Flush reason whenever a flush is scheduled.
Closes https://github.com/facebook/rocksdb/pull/3401

Differential Revision: D6788142

Pulled By: miasantreble

fbshipit-source-id: a867e54d493c06adf5172bd36a180fb3faae3511

Summary:
…db_test

options_settable_test won't pass UBSAN so disable it.
blob_db_test fails in UBSAN as SnapshotList doesn't initialize all the fields in dummy snapshot. Fix it. I don't understand why only blob_db_test fails though.
Closes https://github.com/facebook/rocksdb/pull/3477

Differential Revision: D6928681

Pulled By: siying

fbshipit-source-id: e31dd300fcdecdfd4f6af279a0987fd0cdec5122

Summary:
Update compaction_iterator_test with write-prepared transaction DB related tests. Transaction related tests are group in CompactionIteratorWithSnapshotCheckerTest. The existing test are duplicated to make them also test with dummy SnapshotChecker that will say every key is visible to every snapshot (this is okay, we still compare sequence number to verify visibility). Merge related tests are disabled and will be revisit in another PR.

Existing db_iterator_tests are also duplicated to test with dummy read_callback that will say every key is committed.
Closes https://github.com/facebook/rocksdb/pull/3466

Differential Revision: D6909253

Pulled By: yiwu-arbug

fbshipit-source-id: 2ae4656b843a55e2e9ff8beecf21f2832f96cd25

Summary:
This patch takes advantage of memtable being able to detect duplicate <key,seq> and returning TryAgain to handle duplicate keys in WritePrepared Txns. Through WriteBatchWithIndex's index it detects existence of at least a duplicate key in the write batch. If duplicate key was reported, it then pays the cost of counting the number of sub-patches by iterating over the write batch and pass it to DBImpl::Write. DB will make use of the provided batch_count to assign proper sequence numbers before sending them to the WAL. When later inserting the batch to the memtable, it increases the seq each time memtbale reports a duplicate (a sub-patch in our counting) and tries again.
Closes https://github.com/facebook/rocksdb/pull/3455

Differential Revision: D6873699

Pulled By: maysamyabandeh

fbshipit-source-id: db8487526c3a5dc1ddda0ea49f0f979b26ae648d

Summary:
There are a couple of places where we swallow any error from
WriteBuffer() - in SwitchMemtable() and DBImpl::CloseImpl(). Propagate
the error up in those cases rather than ignoring it.
Closes https://github.com/facebook/rocksdb/pull/3404

Differential Revision: D6879954

Pulled By: anand1976

fbshipit-source-id: 2ef88b554be5286b0a8bad7384ba17a105395bdb

Summary:
ForwardIterator::SVCleanup() sometimes didn't pin superversion when it was supposed to. See the added test for the scenario. Here's the ASAN output of the added test without the fix (using `COMPILE_WITH_ASAN=1 make`): https://pastebin.com/9rD0Ywws
Closes https://github.com/facebook/rocksdb/pull/3415

Differential Revision: D6817414

Pulled By: al13n321

fbshipit-source-id: bc80c44ea78a3a1fa885dfa448a26111f91afb24

Summary:
- Use `__unused__` everywhere
- Suppress unused warnings in Release mode
    + This currently affects non-MSVC builds (e.g. mingw64).
Closes https://github.com/facebook/rocksdb/pull/3448

Differential Revision: D6885496

Pulled By: miasantreble

fbshipit-source-id: f2f6adacec940cc3851a9eee328fafbf61aad211

Summary:
Per some discussions, this will switch our Appveyor testing to use Visual Studio 2017.
Closes https://github.com/facebook/rocksdb/pull/3445

Differential Revision: D6874918

Pulled By: gfosco

fbshipit-source-id: c5a0032ca9f37f0d3baeae35c59d850d528c3176

Summary:
Currently DB does not accept duplicate keys (keys with the same user key and the same sequence number). If Memtable returns false when receiving such keys, we can benefit from this signal to properly increase the sequence number in the rare cases when we have a duplicate key in the write batch written to DB under WritePrepared transactions.
Closes https://github.com/facebook/rocksdb/pull/3418

Differential Revision: D6822412

Pulled By: maysamyabandeh

fbshipit-source-id: adea3ce5073131cd38ed52b16bea0673b1a19e77

Summary:
For #3407
Closes https://github.com/facebook/rocksdb/pull/3425

Differential Revision: D6836900

Pulled By: gfosco

fbshipit-source-id: 7bcaf7a1beeeeabb7c05584f2745e7b4a2473497

Summary:
Updated the test case to handle tmpfs mounted at directories different from "/dev/shm/".
Closes https://github.com/facebook/rocksdb/pull/3440

Differential Revision: D6848213

Pulled By: ajkr

fbshipit-source-id: 465e9dbf0921d0930161f732db6b3766bb030589

Summary:
Using `DeleteFilesInRange` to delete files in a lot of ranges can be slow, because
`VersionSet::LogAndApply` is expensive.

This PR adds a new `DeleteFilesInRange` function to delete files in multiple
ranges at once.

Close https://github.com/facebook/rocksdb/issues/2951
Closes https://github.com/facebook/rocksdb/pull/3431

Differential Revision: D6849228

Pulled By: ajkr

fbshipit-source-id: daeedcabd8def4b1d9ee95a58266dee77b5d68cb

Summary:
In the test, there can be a dead lock between background flush thread and foreground main thread as following:
* background flush thread:
  - holding db mutex, while
  - waiting on "DBImpl::FlushMemTableToOutputFile:BeforeInstallSV" sync point.
* foreground thread:
  - waiting for db mutex to write "key2"

Fixing by let background flush thread wait without holding db mutex.
Closes https://github.com/facebook/rocksdb/pull/3436

Differential Revision: D6841334

Pulled By: yiwu-arbug

fbshipit-source-id: b020768ac94e166e40953c5d09e505515a5f244d

Summary:
Added a test for three dynamic universal compaction options, in the realm of read amplification:
- size_ratio
- min_merge_width
- max_merge_width

Also updated DynamicUniversalCompactionSizeAmplification by adding a check on compaction reason.
Found a bug in compaction reason setting while working on this PR, and fixed in #3412 .

TODO for later: Still to add tests for these options: compression_size_percent, stop_style and trivial_move.
Closes https://github.com/facebook/rocksdb/pull/3419

Differential Revision: D6822217

Pulled By: sagar0

fbshipit-source-id: 074573fca6389053cbac229891a0163f38bb56c4

Summary:
ReadOptions.fill_cache is set in compaction inputs and can be set by users in their queries too. It tells RocksDB not to put a data block used to block cache.

The memory used by the data block is, however, not trackable by users.

To make the system more manageable, we can cost the block to block cache while using it, and then release it after using.
Closes https://github.com/facebook/rocksdb/pull/3333

Differential Revision: D6670230

Pulled By: miasantreble

fbshipit-source-id: ab848d3ed286bd081a13ee1903de357b56cbc308

Summary: Grandfather in super old lint issues to make a clean slate for moving forward that allows us to have stronger enforcement on new issues.

Reviewed By: yiwu-arbug

Differential Revision: D6821806

fbshipit-source-id: 22797d31ec58e9eb0255d3b66fedfcfcb0dc127c

Summary:
While writing tests for dynamic Universal Compaction options, I found that the compaction reasons we set for size-ratio based and sorted-run based universal compactions are swapped with each other. Fixed it.
Closes https://github.com/facebook/rocksdb/pull/3412

Differential Revision: D6820540

Pulled By: sagar0

fbshipit-source-id: 270a188968ba25b2c96a8339904416c4c87ff5b3

Summary:
In DBIter, Prev() calls FindValueForCurrentKey() to search the current value backward. If it finds that there are too many stale value being skipped, it falls back to FindValueForCurrentKeyUsingSeek(), seeking directly to the key with snapshot sequence. After introducing read_callback, however, the key it seeks to might not be visible, according to read_callback. It thus needs to keep searching forward until the first visible value.
Closes https://github.com/facebook/rocksdb/pull/3382

Differential Revision: D6756148

Pulled By: yiwu-arbug

fbshipit-source-id: 064e39b1eec5e083af1c10142600f26d1d2697be

Summary:
In ColumnFamilySet destructor, assert it hold the last reference to cfd before destroy them.

Closes #3112
Closes https://github.com/facebook/rocksdb/pull/3397

Differential Revision: D6777967

Pulled By: yiwu-arbug

fbshipit-source-id: 60b19070e0c194b3b6146699140c1d68777866cb

Summary:
DB::DumpSupportInfo should log all supported compression types.
Closes #3146
Closes https://github.com/facebook/rocksdb/pull/3396

Differential Revision: D6777019

Pulled By: yiwu-arbug

fbshipit-source-id: 5b17f1ffb2d71224e52f7d9c045434746c789fb0

Summary:
This is a small amount of general cleanup I made while experimenting with https://github.com/facebook/rocksdb/issues/3391.
Closes https://github.com/facebook/rocksdb/pull/3392

Differential Revision: D6788365

Pulled By: yiwu-arbug

fbshipit-source-id: 2716e5aabd5424a4dfdaa954361a62c8eb721ae2

Summary:
Now in leveled compaction, we allocate solely based on output target file size. If the total input size is smaller than the number, we should use the total input size instead. Also, cap the allocate size to 1GB.
Closes https://github.com/facebook/rocksdb/pull/3385

Differential Revision: D6762363

Pulled By: siying

fbshipit-source-id: e30906f6e9bff3ec847d2166e44cb49c92f98a13

Summary:
Add an option for SstFileWriter to skip building bloom filters
Closes https://github.com/facebook/rocksdb/pull/3360

Differential Revision: D6709120

Pulled By: IslamAbdelRahman

fbshipit-source-id: 964d4bce38822a048691792f447bcfbb4b6bd809

Summary:
FreeBSD uses jemalloc as the base malloc implementation.
The patch has been functional on FreeBSD as of the MariaDB 10.2 port.
Closes https://github.com/facebook/rocksdb/pull/3386

Differential Revision: D6765742

Pulled By: yiwu-arbug

fbshipit-source-id: d55dbc082eecf640ef3df9a21f26064ebe6587e8

Summary:
Fix data race found by TSAN around WriteStallListener: https://gist.github.com/yiwu-arbug/027d2448b903648f2f0f40b05258d80f
Closes https://github.com/facebook/rocksdb/pull/3384

Differential Revision: D6762167

Pulled By: yiwu-arbug

fbshipit-source-id: cd3a5c9f806de390bd1af6077ea6dbbc8bcaec09

Summary:
Flush() call could be waiting indefinitely if min_write_buffer_number_to_merge is used. Consider the sequence:
1. User call Flush() with flush_options.wait = true
2. The manual flush started in the background
3. New memtable become immutable because of writes. The new memtable will not trigger flush if min_write_buffer_number_to_merge is not reached.
4. The manual flush finish.

Because of the new memtable created at step 3 not being flush, previous logic of WaitForFlushMemTable() keep waiting, despite the memtables it intent to flush has been flushed.

Here instead of checking if there are any more memtables to flush, WaitForFlushMemTable() also check the id of the earliest memtable. If the id is larger than that of latest memtable at the time flush was initiated, it means all the memtable at the time of flush start has all been flush.
Closes https://github.com/facebook/rocksdb/pull/3378

Differential Revision: D6746789

Pulled By: yiwu-arbug

fbshipit-source-id: 35e698f71c7f90b06337a93e6825f4ea3b619bfa

Summary:
Fixes for msys2 and mingw, hide exceptions into cpp  file.
Closes https://github.com/facebook/rocksdb/pull/3377

Differential Revision: D6746707

Pulled By: yiwu-arbug

fbshipit-source-id: 456b38df80bc48b8386a2cf87f669b5a4f9999a4