@nxtedition/rocksdb 5.2.21 → 5.2.26

package/deps/rocksdb/rocksdb/include/rocksdb/filter_policy.h

@@ -28,7 +28,9 @@
 #include <vector>
 
 #include "rocksdb/advanced_options.h"
+#include "rocksdb/customizable.h"
 #include "rocksdb/status.h"
+#include "rocksdb/types.h"
 
 namespace ROCKSDB_NAMESPACE {
 
@@ -36,58 +38,9 @@ class Slice;
 struct BlockBasedTableOptions;
 struct ConfigOptions;
 
-// A class that takes a bunch of keys, then generates filter
-class FilterBitsBuilder {
- public:
-  virtual ~FilterBitsBuilder() {}
-
-  // Add Key to filter, you could use any way to store the key.
-  // Such as: storing hashes or original keys
-  // Keys are in sorted order and duplicated keys are possible.
-  virtual void AddKey(const Slice& key) = 0;
-
-  // Generate the filter using the keys that are added
-  // The return value of this function would be the filter bits,
-  // The ownership of actual data is set to buf
-  virtual Slice Finish(std::unique_ptr<const char[]>* buf) = 0;
-
-  // Approximate the number of keys that can be added and generate a filter
-  // <= the specified number of bytes. Callers (including RocksDB) should
-  // only use this result for optimizing performance and not as a guarantee.
-  // This default implementation is for compatibility with older custom
-  // FilterBitsBuilders only implementing deprecated CalculateNumEntry.
-  virtual size_t ApproximateNumEntries(size_t bytes) {
-    bytes = std::min(bytes, size_t{0xffffffff});
-    return static_cast<size_t>(CalculateNumEntry(static_cast<uint32_t>(bytes)));
-  }
-
-  // Old, DEPRECATED version of ApproximateNumEntries. This is not
-  // called by RocksDB except as the default implementation of
-  // ApproximateNumEntries for API compatibility.
-  virtual int CalculateNumEntry(const uint32_t bytes) {
-    // DEBUG: ideally should not rely on this implementation
-    assert(false);
-    // RELEASE: something reasonably conservative: 2 bytes per entry
-    return static_cast<int>(bytes / 2);
-  }
-};
-
-// A class that checks if a key can be in filter
-// It should be initialized by Slice generated by BitsBuilder
-class FilterBitsReader {
- public:
-  virtual ~FilterBitsReader() {}
-
-  // Check if the entry match the bits in filter
-  virtual bool MayMatch(const Slice& entry) = 0;
-
-  // Check if an array of entries match the bits in filter
-  virtual void MayMatch(int num_keys, Slice** keys, bool* may_match) {
-    for (int i = 0; i < num_keys; ++i) {
-      may_match[i] = MayMatch(*keys[i]);
-    }
-  }
-};
+// As of RocksDB 7.0, the details of these classes are internal
+class FilterBitsBuilder;
+class FilterBitsReader;
 
 // Contextual information passed to BloomFilterPolicy at filter building time.
 // Used in overriding FilterPolicy::GetBuilderWithContext(). References other
@@ -99,100 +52,89 @@ struct FilterBuildingContext {
   // Options for the table being built
   const BlockBasedTableOptions& table_options;
 
-  // Name of the column family for the table (or empty string if unknown)
-  std::string column_family_name;
-
-  // The compactions style in effect for the table
+  // BEGIN from (DB|ColumnFamily)Options in effect at table creation time
   CompactionStyle compaction_style = kCompactionStyleLevel;
 
-  // The table level at time of constructing the SST file, or -1 if unknown.
-  // (The table file could later be used at a different level.)
-  int level_at_creation = -1;
+  // Number of LSM levels, or -1 if unknown
+  int num_levels = -1;
 
   // An optional logger for reporting errors, warnings, etc.
   Logger* info_log = nullptr;
+  // END from (DB|ColumnFamily)Options
+
+  // Name of the column family for the table (or empty string if unknown)
+  // TODO: consider changing to Slice
+  std::string column_family_name;
+
+  // The table level at time of constructing the SST file, or -1 if unknown
+  // or N/A as in SstFileWriter. (The table file could later be used at a
+  // different level.)
+  int level_at_creation = -1;
+
+  // True if known to be going into bottommost sorted run for applicable
+  // key range (which might not even be last level with data). False
+  // otherwise.
+  bool is_bottommost = false;
+
+  // Reason for creating the file with the filter
+  TableFileCreationReason reason = TableFileCreationReason::kMisc;
 };
 
-// We add a new format of filter block called full filter block
-// This new interface gives you more space of customization
-//
-// For the full filter block, you can plug in your version by implement
-// the FilterBitsBuilder and FilterBitsReader
-//
-// There are two sets of interface in FilterPolicy
-// Set 1: CreateFilter, KeyMayMatch: used for blockbased filter
-// Set 2: GetFilterBitsBuilder, GetFilterBitsReader, they are used for
-// full filter.
-// Set 1 MUST be implemented correctly, Set 2 is optional
-// RocksDB would first try using functions in Set 2. if they return nullptr,
-// it would use Set 1 instead.
-// You can choose filter type in NewBloomFilterPolicy
-class FilterPolicy {
+// Determines what kind of filter (if any) to generate in SST files, and under
+// which conditions. API users can create custom filter policies that
+// defer to other built-in policies (see NewBloomFilterPolicy and
+// NewRibbonFilterPolicy) based on the context provided to
+// GetBuilderWithContext.
+class FilterPolicy : public Customizable {
  public:
   virtual ~FilterPolicy();
+  static const char* Type() { return "FilterPolicy"; }
+
+  // The name used for identifying whether a filter on disk is readable
+  // by this FilterPolicy. If this FilterPolicy is part of a family that
+  // can read each others filters, such as built-in BloomFilterPolcy and
+  // RibbonFilterPolicy, the CompatibilityName is a shared family name,
+  // while kinds of filters in the family can have distinct Customizable
+  // Names. This function is pure virtual so that wrappers around built-in
+  // policies are prompted to defer to CompatibilityName() of the wrapped
+  // policy, which is important for compatibility.
+  //
+  // For custom filter policies that are not part of a read-compatible
+  // family (rare), implementations may return Name().
+  virtual const char* CompatibilityName() const = 0;
 
   // Creates a new FilterPolicy based on the input value string and returns the
   // result The value might be an ID, and ID with properties, or an old-style
   // policy string.
   // The value describes the FilterPolicy being created.
   // For BloomFilters, value may be a ":"-delimited value of the form:
-  //   "bloomfilter:[bits_per_key]:[use_block_based_builder]",
-  //   e.g. ""bloomfilter:4:true"
-  //   The above string is equivalent to calling NewBloomFilterPolicy(4, true).
+  //   "bloomfilter:[bits_per_key]",
+  //   e.g. ""bloomfilter:4"
+  //   The above string is equivalent to calling NewBloomFilterPolicy(4).
   static Status CreateFromString(const ConfigOptions& config_options,
                                  const std::string& value,
                                  std::shared_ptr<const FilterPolicy>* result);
 
-  // Return the name of this policy.  Note that if the filter encoding
-  // changes in an incompatible way, the name returned by this method
-  // must be changed.  Otherwise, old incompatible filters may be
-  // passed to methods of this type.
-  virtual const char* Name() const = 0;
-
-  // keys[0,n-1] contains a list of keys (potentially with duplicates)
-  // that are ordered according to the user supplied comparator.
-  // Append a filter that summarizes keys[0,n-1] to *dst.
-  //
-  // Warning: do not change the initial contents of *dst.  Instead,
-  // append the newly constructed filter to *dst.
-  virtual void CreateFilter(const Slice* keys, int n,
-                            std::string* dst) const = 0;
-
-  // "filter" contains the data appended by a preceding call to
-  // CreateFilter() on this class.  This method must return true if
-  // the key was in the list of keys passed to CreateFilter().
-  // This method may return true or false if the key was not on the
-  // list, but it should aim to return false with a high probability.
-  virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const = 0;
-
-  // Return a new FilterBitsBuilder for full or partitioned filter blocks, or
-  // nullptr if using block-based filter.
-  // NOTE: This function is only called by GetBuilderWithContext() below for
-  // custom FilterPolicy implementations. Thus, it is not necessary to
-  // override this function if overriding GetBuilderWithContext().
-  virtual FilterBitsBuilder* GetFilterBitsBuilder() const { return nullptr; }
-
-  // A newer variant of GetFilterBitsBuilder that allows a FilterPolicy
-  // to customize the builder for contextual constraints and hints.
-  // (Name changed to avoid triggering -Werror=overloaded-virtual.)
-  // If overriding GetFilterBitsBuilder() suffices, it is not necessary to
-  // override this function.
+  // Return a new FilterBitsBuilder for constructing full or partitioned
+  // filter blocks, or return nullptr to indicate "no filter". Custom
+  // implementations should defer to a built-in FilterPolicy to get a
+  // new FilterBitsBuilder, but the FilterBuildingContext can be used
+  // to decide which built-in FilterPolicy to defer to.
   virtual FilterBitsBuilder* GetBuilderWithContext(
-      const FilterBuildingContext&) const {
-    return GetFilterBitsBuilder();
-  }
+      const FilterBuildingContext&) const = 0;
 
-  // Return a new FilterBitsReader for full or partitioned filter blocks, or
-  // nullptr if using block-based filter.
-  // As here, the input slice should NOT be deleted by FilterPolicy.
+  // Return a new FilterBitsReader for full or partitioned filter blocks.
+  // Caller retains ownership of any buffer pointed to by the input Slice.
+  // Custom implementation should defer to GetFilterBitsReader on any
+  // built-in FilterPolicy, which can read filters generated by any other
+  // built-in FilterPolicy.
   virtual FilterBitsReader* GetFilterBitsReader(
-      const Slice& /*contents*/) const {
-    return nullptr;
-  }
+      const Slice& /*contents*/) const = 0;
 };
 
 // Return a new filter policy that uses a bloom filter with approximately
-// the specified number of bits per key.
+// the specified number of bits per key. See
+// https://github.com/facebook/rocksdb/wiki/RocksDB-Bloom-Filter
 //
 // bits_per_key: average bits allocated per key in bloom filter. A good
 // choice is 9.9, which yields a filter with ~ 1% false positive rate.
@@ -200,11 +142,18 @@ class FilterPolicy {
 // integer. Recommend using no more than three decimal digits after the
 // decimal point, as in 6.667.
 //
-// use_block_based_builder: use deprecated block based filter (true) rather
-// than full or partitioned filter (false).
+// To avoid configurations that are unlikely to produce good filtering
+// value for the CPU overhead, bits_per_key < 0.5 is rounded down to 0.0
+// which means "generate no filter", and 0.5 <= bits_per_key < 1.0 is
+// rounded up to 1.0, for a 62% FP rate.
+//
+// The caller is responsible for eventually deleting the result, though
+// this is typically handled automatically with BlockBasedTableOptions:
+//   table_options.filter_policy.reset(NewBloomFilterPolicy(...));
 //
-// Callers must delete the result after any database that is using the
-// result has been closed.
+// As of RocksDB 7.0, the use_block_based_builder parameter is ignored.
+// (The old, inefficient block-based filter is no longer accessible in
+// the public API.)
 //
 // Note: if you are using a custom comparator that ignores some parts
 // of the keys being compared, you must not use NewBloomFilterPolicy()
@@ -214,25 +163,44 @@ class FilterPolicy {
 // FilterPolicy (like NewBloomFilterPolicy) that does not ignore
 // trailing spaces in keys.
 extern const FilterPolicy* NewBloomFilterPolicy(
-    double bits_per_key, bool use_block_based_builder = false);
-
-// An EXPERIMENTAL new Bloom alternative that saves about 30% space
-// compared to Bloom filters, with about 3-4x construction time and
-// similar query times. For example, if you pass in 10 for
-// bloom_equivalent_bits_per_key, you'll get the same 0.95% FP rate
-// as Bloom filter but only using about 7 bits per key. (This
-// way of configuring the new filter is considered experimental
-// and/or transitional, so is expected to go away.)
+    double bits_per_key, bool IGNORED_use_block_based_builder = false);
+
+// A new Bloom alternative that saves about 30% space compared to
+// Bloom filters, with similar query times but roughly 3-4x CPU time
+// and 3x temporary space usage during construction.  For example, if
+// you pass in 10 for bloom_equivalent_bits_per_key, you'll get the same
+// 0.95% FP rate as Bloom filter but only using about 7 bits per key.
+//
+// The space savings of Ribbon filters makes sense for lower (higher
+// numbered; larger; longer-lived) levels of LSM, whereas the speed of
+// Bloom filters make sense for highest levels of LSM. Setting
+// bloom_before_level allows for this design with Level and Universal
+// compaction styles. For example, bloom_before_level=1 means that Bloom
+// filters will be used in level 0, including flushes, and Ribbon
+// filters elsewhere, including FIFO compaction and external SST files.
+// For this option, memtable flushes are considered level -1 (so that
+// flushes can be distinguished from intra-L0 compaction).
+// bloom_before_level=0 (default) -> Generate Bloom filters only for
+// flushes under Level and Universal compaction styles.
+// bloom_before_level=-1 -> Always generate Ribbon filters (except in
+// some extreme or exceptional cases).
+//
+// Ribbon filters are compatible with RocksDB >= 6.15.0. Earlier
+// versions reading the data will behave as if no filter was used
+// (degraded performance until compaction rebuilds filters). All
+// built-in FilterPolicies (Bloom or Ribbon) are able to read other
+// kinds of built-in filters.
 //
-// Ribbon filters are ignored by previous versions of RocksDB, as if
-// no filter was used.
+// Note: the current Ribbon filter schema uses some extra resources
+// when constructing very large filters. For example, for 100 million
+// keys in a single filter (one SST file without partitioned filters),
+// 3GB of temporary, untracked memory is used, vs. 1GB for Bloom.
+// However, the savings in filter space from just ~60 open SST files
+// makes up for the additional temporary memory use.
 //
-// Note: this policy can generate Bloom filters in some cases.
-// For very small filters (well under 1KB), Bloom fallback is by
-// design, as the current Ribbon schema is not optimized to save vs.
-// Bloom for such small filters. Other cases of Bloom fallback should
-// be exceptional and log an appropriate warning.
-extern const FilterPolicy* NewExperimentalRibbonFilterPolicy(
-    double bloom_equivalent_bits_per_key);
+// Also consider using optimize_filters_for_memory to save filter
+// memory.
+extern const FilterPolicy* NewRibbonFilterPolicy(
+    double bloom_equivalent_bits_per_key, int bloom_before_level = 0);
 
 }  // namespace ROCKSDB_NAMESPACE

package/deps/rocksdb/rocksdb/include/rocksdb/flush_block_policy.h

@@ -6,16 +6,23 @@
 #pragma once
 
 #include <string>
+
+#include "rocksdb/customizable.h"
 #include "rocksdb/table.h"
 
 namespace ROCKSDB_NAMESPACE {
 
 class Slice;
 class BlockBuilder;
+struct ConfigOptions;
 struct Options;
 
 // FlushBlockPolicy provides a configurable way to determine when to flush a
-// block in the block based tables,
+// block in the block based tables.
+//
+// Exceptions MUST NOT propagate out of overridden functions into RocksDB,
+// because RocksDB is not exception-safe. This could cause undefined behavior
+// including data loss, unreported corruption, deadlocks, and more.
 class FlushBlockPolicy {
  public:
   // Keep track of the key/value sequences and return the boolean value to
@@ -25,10 +32,16 @@ class FlushBlockPolicy {
   virtual ~FlushBlockPolicy() {}
 };
 
-class FlushBlockPolicyFactory {
+class FlushBlockPolicyFactory : public Customizable {
  public:
-  // Return the name of the flush block policy.
-  virtual const char* Name() const = 0;
+  static const char* Type() { return "FlushBlockPolicyFactory"; }
+
+  // Creates a FlushBlockPolicyFactory based on the input value.
+  // By default, this method can create EveryKey or BySize PolicyFactory,
+  // which take now config_options.
+  static Status CreateFromString(
+      const ConfigOptions& config_options, const std::string& value,
+      std::shared_ptr<FlushBlockPolicyFactory>* result);
 
   // Return a new block flush policy that flushes data blocks by data size.
   // FlushBlockPolicy may need to access the metadata of the data block
@@ -45,9 +58,10 @@ class FlushBlockPolicyFactory {
 
 class FlushBlockBySizePolicyFactory : public FlushBlockPolicyFactory {
  public:
-  FlushBlockBySizePolicyFactory() {}
+  FlushBlockBySizePolicyFactory();
 
-  const char* Name() const override { return "FlushBlockBySizePolicyFactory"; }
+  static const char* kClassName() { return "FlushBlockBySizePolicyFactory"; }
+  const char* Name() const override { return kClassName(); }
 
   FlushBlockPolicy* NewFlushBlockPolicy(
       const BlockBasedTableOptions& table_options,

package/deps/rocksdb/rocksdb/include/rocksdb/functor_wrapper.h

@@ -0,0 +1,56 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include <functional>
+#include <memory>
+#include <utility>
+
+#include "rocksdb/rocksdb_namespace.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+namespace detail {
+template <std::size_t...>
+struct IndexSequence {};
+
+template <std::size_t N, std::size_t... Next>
+struct IndexSequenceHelper
+    : public IndexSequenceHelper<N - 1U, N - 1U, Next...> {};
+
+template <std::size_t... Next>
+struct IndexSequenceHelper<0U, Next...> {
+  using type = IndexSequence<Next...>;
+};
+
+template <std::size_t N>
+using make_index_sequence = typename IndexSequenceHelper<N>::type;
+
+template <typename Function, typename Tuple, size_t... I>
+void call(Function f, Tuple t, IndexSequence<I...>) {
+  f(std::get<I>(t)...);
+}
+
+template <typename Function, typename Tuple>
+void call(Function f, Tuple t) {
+  static constexpr auto size = std::tuple_size<Tuple>::value;
+  call(f, t, make_index_sequence<size>{});
+}
+}  // namespace detail
+
+template <typename... Args>
+class FunctorWrapper {
+ public:
+  explicit FunctorWrapper(std::function<void(Args...)> functor, Args &&...args)
+      : functor_(std::move(functor)), args_(std::forward<Args>(args)...) {}
+
+  void invoke() { detail::call(functor_, args_); }
+
+ private:
+  std::function<void(Args...)> functor_;
+  std::tuple<Args...> args_;
+};
+}  // namespace ROCKSDB_NAMESPACE

package/deps/rocksdb/rocksdb/include/rocksdb/io_status.h

@@ -28,7 +28,7 @@ class IOStatus : public Status {
   using Code = Status::Code;
   using SubCode = Status::SubCode;
 
-  enum IOErrorScope {
+  enum IOErrorScope : unsigned char {
     kIOErrorScopeFileSystem,
     kIOErrorScopeFile,
     kIOErrorScopeRange,
@@ -57,11 +57,13 @@ class IOStatus : public Status {
 
   void SetRetryable(bool retryable) { retryable_ = retryable; }
   void SetDataLoss(bool data_loss) { data_loss_ = data_loss; }
-  void SetScope(IOErrorScope scope) { scope_ = scope; }
+  void SetScope(IOErrorScope scope) {
+    scope_ = static_cast<unsigned char>(scope);
+  }
 
   bool GetRetryable() const { return retryable_; }
   bool GetDataLoss() const { return data_loss_; }
-  IOErrorScope GetScope() const { return scope_; }
+  IOErrorScope GetScope() const { return static_cast<IOErrorScope>(scope_); }
 
   // Return a success status.
   static IOStatus OK() { return IOStatus(); }
@@ -137,15 +139,9 @@ class IOStatus : public Status {
 
  private:
   friend IOStatus status_to_io_status(Status&&);
-  bool retryable_;
-  bool data_loss_;
-  IOErrorScope scope_;
 
   explicit IOStatus(Code _code, SubCode _subcode = kNone)
-      : Status(_code, _subcode),
-        retryable_(false),
-        data_loss_(false),
-        scope_(kIOErrorScopeFileSystem) {}
+      : Status(_code, _subcode, false, false, kIOErrorScopeFileSystem) {}
 
   IOStatus(Code _code, SubCode _subcode, const Slice& msg, const Slice& msg2);
   IOStatus(Code _code, const Slice& msg, const Slice& msg2)
@@ -154,10 +150,7 @@ class IOStatus : public Status {
 
 inline IOStatus::IOStatus(Code _code, SubCode _subcode, const Slice& msg,
                           const Slice& msg2)
-    : Status(_code, _subcode),
-      retryable_(false),
-      data_loss_(false),
-      scope_(kIOErrorScopeFileSystem) {
+    : Status(_code, _subcode, false, false, kIOErrorScopeFileSystem) {
   assert(code_ != kOk);
   assert(subcode_ != kMaxSubCode);
   const size_t len1 = msg.size();
@@ -171,7 +164,7 @@ inline IOStatus::IOStatus(Code _code, SubCode _subcode, const Slice& msg,
     memcpy(result + len1 + 2, msg2.data(), len2);
   }
   result[size] = '\0';  // null terminator for C style string
-  state_ = result;
+  state_.reset(result);
 }
 
 inline IOStatus::IOStatus(const IOStatus& s) : Status(s.code_, s.subcode_) {
@@ -181,7 +174,7 @@ inline IOStatus::IOStatus(const IOStatus& s) : Status(s.code_, s.subcode_) {
   retryable_ = s.retryable_;
   data_loss_ = s.data_loss_;
   scope_ = s.scope_;
-  state_ = (s.state_ == nullptr) ? nullptr : CopyState(s.state_);
+  state_ = (s.state_ == nullptr) ? nullptr : CopyState(s.state_.get());
 }
 inline IOStatus& IOStatus::operator=(const IOStatus& s) {
   // The following condition catches both aliasing (when this == &s),
@@ -196,8 +189,7 @@ inline IOStatus& IOStatus::operator=(const IOStatus& s) {
     retryable_ = s.retryable_;
     data_loss_ = s.data_loss_;
     scope_ = s.scope_;
-    delete[] state_;
-    state_ = (s.state_ == nullptr) ? nullptr : CopyState(s.state_);
+    state_ = (s.state_ == nullptr) ? nullptr : CopyState(s.state_.get());
   }
   return *this;
 }
@@ -228,9 +220,7 @@ inline IOStatus& IOStatus::operator=(IOStatus&& s)
     data_loss_ = s.data_loss_;
     scope_ = s.scope_;
     s.scope_ = kIOErrorScopeFileSystem;
-    delete[] state_;
-    state_ = nullptr;
-    std::swap(state_, s.state_);
+    state_ = std::move(s.state_);
   }
   return *this;
 }
@@ -252,18 +242,10 @@ inline bool IOStatus::operator!=(const IOStatus& rhs) const {
 }
 
 inline IOStatus status_to_io_status(Status&& status) {
-  if (status.ok()) {
-    // Fast path
-    return IOStatus::OK();
-  } else {
-    const char* state = status.getState();
-    if (state) {
-      return IOStatus(status.code(), status.subcode(),
-                      Slice(state, strlen(status.getState()) + 1), Slice());
-    } else {
-      return IOStatus(status.code(), status.subcode());
-    }
-  }
+  IOStatus io_s;
+  Status& s = io_s;
+  s = std::move(status);
+  return io_s;
 }
 
 }  // namespace ROCKSDB_NAMESPACE

package/deps/rocksdb/rocksdb/include/rocksdb/iostats_context.h

@@ -14,6 +14,32 @@
 
 namespace ROCKSDB_NAMESPACE {
 
+// EXPERIMENTAL: the IO statistics for tiered storage. It matches with each
+// item in Temperature class.
+struct FileIOByTemperature {
+  // the number of bytes read to Temperature::kHot file
+  uint64_t hot_file_bytes_read;
+  // the number of bytes read to Temperature::kWarm file
+  uint64_t warm_file_bytes_read;
+  // the number of bytes read to Temperature::kCold file
+  uint64_t cold_file_bytes_read;
+  // total number of reads to Temperature::kHot file
+  uint64_t hot_file_read_count;
+  // total number of reads to Temperature::kWarm file
+  uint64_t warm_file_read_count;
+  // total number of reads to Temperature::kCold file
+  uint64_t cold_file_read_count;
+  // reset all the statistics to 0.
+  void Reset() {
+    hot_file_bytes_read = 0;
+    warm_file_bytes_read = 0;
+    cold_file_bytes_read = 0;
+    hot_file_read_count = 0;
+    warm_file_read_count = 0;
+    cold_file_read_count = 0;
+  }
+};
+
 struct IOStatsContext {
   // reset all io-stats counter to zero
   void Reset();
@@ -48,9 +74,19 @@ struct IOStatsContext {
   uint64_t cpu_write_nanos;
   // CPU time spent in read() and pread()
   uint64_t cpu_read_nanos;
+
+  FileIOByTemperature file_io_stats_by_temperature;
 };
 
-// Get Thread-local IOStatsContext object pointer
+// If RocksDB is compiled with -DNIOSTATS_CONTEXT, then a pointer to a global,
+// non-thread-local IOStatsContext object will be returned. Attempts to update
+// this object will be ignored, and reading from it will also be no-op.
+// Otherwise,
+// a) if thread-local is supported on the platform, then a pointer to
+//    a thread-local IOStatsContext object will be returned.
+// b) if thread-local is NOT supported, then compilation will fail.
+//
+// This function never returns nullptr.
 IOStatsContext* get_iostats_context();
 
 }  // namespace ROCKSDB_NAMESPACE

package/deps/rocksdb/rocksdb/include/rocksdb/iterator.h

@@ -45,7 +45,6 @@ class Iterator : public Cleanable {
 
   // Position at the last key in the source.  The iterator is
   // Valid() after this call iff the source is not empty.
-  // Currently incompatible with user timestamp.
   virtual void SeekToLast() = 0;
 
   // Position at the first key in the source that at or past target.
@@ -60,7 +59,7 @@ class Iterator : public Cleanable {
   // Position at the last key in the source that at or before target.
   // The iterator is Valid() after this call iff the source contains
   // an entry that comes at or before target.
-  // Currently incompatible with user timestamp.
+  // Target does not contain timestamp.
   virtual void SeekForPrev(const Slice& target) = 0;
 
   // Moves to the next entry in the source.  After this call, Valid() is
@@ -70,7 +69,6 @@ class Iterator : public Cleanable {
 
   // Moves to the previous entry in the source.  After this call, Valid() is
   // true iff the iterator was not positioned at the first entry in source.
-  // Currently incompatible with user timestamp.
   // REQUIRES: Valid()
   virtual void Prev() = 0;