@nxtedition/rocksdb 5.2.21 → 5.2.26

package/deps/rocksdb/rocksdb/file/random_access_file_reader.cc

@@ -12,6 +12,7 @@
 #include <algorithm>
 #include <mutex>
 
+#include "file/file_util.h"
 #include "monitoring/histogram.h"
 #include "monitoring/iostats_context_imp.h"
 #include "port/port.h"
@@ -22,17 +23,77 @@
 
 namespace ROCKSDB_NAMESPACE {
 
-Status RandomAccessFileReader::Read(const IOOptions& opts, uint64_t offset,
-                                    size_t n, Slice* result, char* scratch,
-                                    AlignedBuf* aligned_buf,
-                                    bool for_compaction) const {
+inline void RecordIOStats(Statistics* stats, Temperature file_temperature,
+                          bool is_last_level, size_t size) {
+  IOSTATS_ADD(bytes_read, size);
+  // record for last/non-last level
+  if (is_last_level) {
+    RecordTick(stats, LAST_LEVEL_READ_BYTES, size);
+    RecordTick(stats, LAST_LEVEL_READ_COUNT, 1);
+  } else {
+    RecordTick(stats, NON_LAST_LEVEL_READ_BYTES, size);
+    RecordTick(stats, NON_LAST_LEVEL_READ_COUNT, 1);
+  }
+
+  // record for temperature file
+  if (file_temperature != Temperature::kUnknown) {
+    switch (file_temperature) {
+      case Temperature::kHot:
+        IOSTATS_ADD(file_io_stats_by_temperature.hot_file_bytes_read, size);
+        IOSTATS_ADD(file_io_stats_by_temperature.hot_file_read_count, 1);
+        RecordTick(stats, HOT_FILE_READ_BYTES, size);
+        RecordTick(stats, HOT_FILE_READ_COUNT, 1);
+        break;
+      case Temperature::kWarm:
+        IOSTATS_ADD(file_io_stats_by_temperature.warm_file_bytes_read, size);
+        IOSTATS_ADD(file_io_stats_by_temperature.warm_file_read_count, 1);
+        RecordTick(stats, WARM_FILE_READ_BYTES, size);
+        RecordTick(stats, WARM_FILE_READ_COUNT, 1);
+        break;
+      case Temperature::kCold:
+        IOSTATS_ADD(file_io_stats_by_temperature.cold_file_bytes_read, size);
+        IOSTATS_ADD(file_io_stats_by_temperature.cold_file_read_count, 1);
+        RecordTick(stats, COLD_FILE_READ_BYTES, size);
+        RecordTick(stats, COLD_FILE_READ_COUNT, 1);
+        break;
+      default:
+        break;
+    }
+  }
+}
+
+IOStatus RandomAccessFileReader::Create(
+    const std::shared_ptr<FileSystem>& fs, const std::string& fname,
+    const FileOptions& file_opts,
+    std::unique_ptr<RandomAccessFileReader>* reader, IODebugContext* dbg) {
+  std::unique_ptr<FSRandomAccessFile> file;
+  IOStatus io_s = fs->NewRandomAccessFile(fname, file_opts, &file, dbg);
+  if (io_s.ok()) {
+    reader->reset(new RandomAccessFileReader(std::move(file), fname));
+  }
+  return io_s;
+}
+
+IOStatus RandomAccessFileReader::Read(
+    const IOOptions& opts, uint64_t offset, size_t n, Slice* result,
+    char* scratch, AlignedBuf* aligned_buf,
+    Env::IOPriority rate_limiter_priority) const {
   (void)aligned_buf;
 
   TEST_SYNC_POINT_CALLBACK("RandomAccessFileReader::Read", nullptr);
-  Status s;
+
+  // To be paranoid: modify scratch a little bit, so in case underlying
+  // FileSystem doesn't fill the buffer but return success and `scratch` returns
+  // contains a previous block, returned value will not pass checksum.
+  if (n > 0 && scratch != nullptr) {
+    // This byte might not change anything for direct I/O case, but it's OK.
+    scratch[0]++;
+  }
+
+  IOStatus io_s;
   uint64_t elapsed = 0;
   {
-    StopWatch sw(env_, stats_, hist_type_,
+    StopWatch sw(clock_, stats_, hist_type_,
                  (stats_ != nullptr) ? &elapsed : nullptr, true /*overwrite*/,
                  true /*delay_enabled*/);
     auto prev_perf_level = GetPerfLevel();
@@ -50,10 +111,11 @@ Status RandomAccessFileReader::Read(const IOOptions& opts, uint64_t offset,
       buf.AllocateNewBuffer(read_size);
       while (buf.CurrentSize() < read_size) {
         size_t allowed;
-        if (for_compaction && rate_limiter_ != nullptr) {
+        if (rate_limiter_priority != Env::IO_TOTAL &&
+            rate_limiter_ != nullptr) {
           allowed = rate_limiter_->RequestToken(
               buf.Capacity() - buf.CurrentSize(), buf.Alignment(),
-              Env::IOPriority::IO_LOW, stats_, RateLimiter::OpType::kRead);
+              rate_limiter_priority, stats_, RateLimiter::OpType::kRead);
         } else {
           assert(buf.CurrentSize() == 0);
           allowed = read_size;
@@ -68,28 +130,32 @@ Status RandomAccessFileReader::Read(const IOOptions& opts, uint64_t offset,
         }
 
         {
-          IOSTATS_CPU_TIMER_GUARD(cpu_read_nanos, env_);
+          IOSTATS_CPU_TIMER_GUARD(cpu_read_nanos, clock_);
           // Only user reads are expected to specify a timeout. And user reads
           // are not subjected to rate_limiter and should go through only
           // one iteration of this loop, so we don't need to check and adjust
           // the opts.timeout before calling file_->Read
           assert(!opts.timeout.count() || allowed == read_size);
-          s = file_->Read(aligned_offset + buf.CurrentSize(), allowed, opts,
-                          &tmp, buf.Destination(), nullptr);
+          io_s = file_->Read(aligned_offset + buf.CurrentSize(), allowed, opts,
+                             &tmp, buf.Destination(), nullptr);
         }
         if (ShouldNotifyListeners()) {
           auto finish_ts = FileOperationInfo::FinishNow();
           NotifyOnFileReadFinish(orig_offset, tmp.size(), start_ts, finish_ts,
-                                 s);
+                                 io_s);
+          if (!io_s.ok()) {
+            NotifyOnIOError(io_s, FileOperationType::kRead, file_name(),
+                            tmp.size(), orig_offset);
+          }
         }
 
         buf.Size(buf.CurrentSize() + tmp.size());
-        if (!s.ok() || tmp.size() < allowed) {
+        if (!io_s.ok() || tmp.size() < allowed) {
           break;
         }
       }
       size_t res_len = 0;
-      if (s.ok() && offset_advance < buf.CurrentSize()) {
+      if (io_s.ok() && offset_advance < buf.CurrentSize()) {
         res_len = std::min(buf.CurrentSize() - offset_advance, n);
         if (aligned_buf == nullptr) {
           buf.Read(scratch, offset_advance, res_len);
@@ -105,12 +171,13 @@ Status RandomAccessFileReader::Read(const IOOptions& opts, uint64_t offset,
       const char* res_scratch = nullptr;
       while (pos < n) {
         size_t allowed;
-        if (for_compaction && rate_limiter_ != nullptr) {
+        if (rate_limiter_priority != Env::IO_TOTAL &&
+            rate_limiter_ != nullptr) {
           if (rate_limiter_->IsRateLimited(RateLimiter::OpType::kRead)) {
             sw.DelayStart();
           }
           allowed = rate_limiter_->RequestToken(n - pos, 0 /* alignment */,
-                                                Env::IOPriority::IO_LOW, stats_,
+                                                rate_limiter_priority, stats_,
                                                 RateLimiter::OpType::kRead);
           if (rate_limiter_->IsRateLimited(RateLimiter::OpType::kRead)) {
             sw.DelayStop();
@@ -128,23 +195,27 @@ Status RandomAccessFileReader::Read(const IOOptions& opts, uint64_t offset,
 #endif
 
         {
-          IOSTATS_CPU_TIMER_GUARD(cpu_read_nanos, env_);
+          IOSTATS_CPU_TIMER_GUARD(cpu_read_nanos, clock_);
           // Only user reads are expected to specify a timeout. And user reads
           // are not subjected to rate_limiter and should go through only
           // one iteration of this loop, so we don't need to check and adjust
           // the opts.timeout before calling file_->Read
           assert(!opts.timeout.count() || allowed == n);
-          s = file_->Read(offset + pos, allowed, opts, &tmp_result,
-                          scratch + pos, nullptr);
+          io_s = file_->Read(offset + pos, allowed, opts, &tmp_result,
+                             scratch + pos, nullptr);
         }
 #ifndef ROCKSDB_LITE
         if (ShouldNotifyListeners()) {
           auto finish_ts = FileOperationInfo::FinishNow();
           NotifyOnFileReadFinish(offset + pos, tmp_result.size(), start_ts,
-                                 finish_ts, s);
+                                 finish_ts, io_s);
+
+          if (!io_s.ok()) {
+            NotifyOnIOError(io_s, FileOperationType::kRead, file_name(),
+                            tmp_result.size(), offset + pos);
+          }
         }
 #endif
-
         if (res_scratch == nullptr) {
           // we can't simply use `scratch` because reads of mmap'd files return
           // data in a different buffer.
@@ -154,20 +225,20 @@ Status RandomAccessFileReader::Read(const IOOptions& opts, uint64_t offset,
           assert(tmp_result.data() == res_scratch + pos);
         }
         pos += tmp_result.size();
-        if (!s.ok() || tmp_result.size() < allowed) {
+        if (!io_s.ok() || tmp_result.size() < allowed) {
           break;
         }
       }
-      *result = Slice(res_scratch, s.ok() ? pos : 0);
+      *result = Slice(res_scratch, io_s.ok() ? pos : 0);
     }
-    IOSTATS_ADD_IF_POSITIVE(bytes_read, result->size());
+    RecordIOStats(stats_, file_temperature_, is_last_level_, result->size());
     SetPerfLevel(prev_perf_level);
   }
   if (stats_ != nullptr && file_read_hist_ != nullptr) {
     file_read_hist_->Add(elapsed);
   }
 
-  return s;
+  return io_s;
 }
 
 size_t End(const FSReadRequest& r) {
@@ -196,16 +267,36 @@ bool TryMerge(FSReadRequest* dest, const FSReadRequest& src) {
   return true;
 }
 
-Status RandomAccessFileReader::MultiRead(const IOOptions& opts,
-                                         FSReadRequest* read_reqs,
-                                         size_t num_reqs,
-                                         AlignedBuf* aligned_buf) const {
+IOStatus RandomAccessFileReader::MultiRead(
+    const IOOptions& opts, FSReadRequest* read_reqs, size_t num_reqs,
+    AlignedBuf* aligned_buf, Env::IOPriority rate_limiter_priority) const {
+  if (rate_limiter_priority != Env::IO_TOTAL) {
+    return IOStatus::NotSupported("Unable to rate limit MultiRead()");
+  }
   (void)aligned_buf;  // suppress warning of unused variable in LITE mode
   assert(num_reqs > 0);
-  Status s;
+
+#ifndef NDEBUG
+  for (size_t i = 0; i < num_reqs - 1; ++i) {
+    assert(read_reqs[i].offset <= read_reqs[i + 1].offset);
+  }
+#endif  // !NDEBUG
+
+  // To be paranoid modify scratch a little bit, so in case underlying
+  // FileSystem doesn't fill the buffer but return succee and `scratch` returns
+  // contains a previous block, returned value will not pass checksum.
+  // This byte might not change anything for direct I/O case, but it's OK.
+  for (size_t i = 0; i < num_reqs; i++) {
+    FSReadRequest& r = read_reqs[i];
+    if (r.len > 0 && r.scratch != nullptr) {
+      r.scratch[0]++;
+    }
+  }
+
+  IOStatus io_s;
   uint64_t elapsed = 0;
   {
-    StopWatch sw(env_, stats_, hist_type_,
+    StopWatch sw(clock_, stats_, hist_type_,
                  (stats_ != nullptr) ? &elapsed : nullptr, true /*overwrite*/,
                  true /*delay_enabled*/);
     auto prev_perf_level = GetPerfLevel();
@@ -267,8 +358,8 @@ Status RandomAccessFileReader::MultiRead(const IOOptions& opts,
 #endif  // ROCKSDB_LITE
 
     {
-      IOSTATS_CPU_TIMER_GUARD(cpu_read_nanos, env_);
-      s = file_->MultiRead(fs_reqs, num_fs_reqs, opts, nullptr);
+      IOSTATS_CPU_TIMER_GUARD(cpu_read_nanos, clock_);
+      io_s = file_->MultiRead(fs_reqs, num_fs_reqs, opts, nullptr);
     }
 
 #ifndef ROCKSDB_LITE
@@ -284,8 +375,14 @@ Status RandomAccessFileReader::MultiRead(const IOOptions& opts,
         r.status = fs_r.status;
         if (r.status.ok()) {
           uint64_t offset = r.offset - fs_r.offset;
-          size_t len = std::min(r.len, static_cast<size_t>(fs_r.len - offset));
-          r.result = Slice(fs_r.scratch + offset, len);
+          if (fs_r.result.size() <= offset) {
+            // No byte in the read range is returned.
+            r.result = Slice();
+          } else {
+            size_t len = std::min(
+                r.len, static_cast<size_t>(fs_r.result.size() - offset));
+            r.result = Slice(fs_r.scratch + offset, len);
+          }
         } else {
           r.result = Slice();
         }
@@ -300,8 +397,15 @@ Status RandomAccessFileReader::MultiRead(const IOOptions& opts,
         NotifyOnFileReadFinish(read_reqs[i].offset, read_reqs[i].result.size(),
                                start_ts, finish_ts, read_reqs[i].status);
       }
+      if (!read_reqs[i].status.ok()) {
+        NotifyOnIOError(read_reqs[i].status, FileOperationType::kRead,
+                        file_name(), read_reqs[i].result.size(),
+                        read_reqs[i].offset);
+      }
+
 #endif  // ROCKSDB_LITE
-      IOSTATS_ADD_IF_POSITIVE(bytes_read, read_reqs[i].result.size());
+      RecordIOStats(stats_, file_temperature_, is_last_level_,
+                    read_reqs[i].result.size());
     }
     SetPerfLevel(prev_perf_level);
   }
@@ -309,7 +413,87 @@ Status RandomAccessFileReader::MultiRead(const IOOptions& opts,
     file_read_hist_->Add(elapsed);
   }
 
+  return io_s;
+}
+
+IOStatus RandomAccessFileReader::PrepareIOOptions(const ReadOptions& ro,
+                                                  IOOptions& opts) {
+  if (clock_ != nullptr) {
+    return PrepareIOFromReadOptions(ro, clock_, opts);
+  } else {
+    return PrepareIOFromReadOptions(ro, SystemClock::Default().get(), opts);
+  }
+}
+
+// TODO akanksha:
+// 1. Handle use_direct_io case which currently calls Read API.
+IOStatus RandomAccessFileReader::ReadAsync(
+    FSReadRequest& req, const IOOptions& opts,
+    std::function<void(const FSReadRequest&, void*)> cb, void* cb_arg,
+    void** io_handle, IOHandleDeleter* del_fn,
+    Env::IOPriority rate_limiter_priority) {
+  if (use_direct_io()) {
+    // For direct_io, it calls Read API.
+    req.status = Read(opts, req.offset, req.len, &(req.result), req.scratch,
+                      nullptr /*dbg*/, rate_limiter_priority);
+    cb(req, cb_arg);
+    return IOStatus::OK();
+  }
+
+  // Create a callback and populate info.
+  auto read_async_callback =
+      std::bind(&RandomAccessFileReader::ReadAsyncCallback, this,
+                std::placeholders::_1, std::placeholders::_2);
+  ReadAsyncInfo* read_async_info = new ReadAsyncInfo;
+  read_async_info->cb_ = cb;
+  read_async_info->cb_arg_ = cb_arg;
+  read_async_info->start_time_ = clock_->NowMicros();
+
+#ifndef ROCKSDB_LITE
+  if (ShouldNotifyListeners()) {
+    read_async_info->fs_start_ts_ = FileOperationInfo::StartNow();
+  }
+#endif
+
+  IOStatus s = file_->ReadAsync(req, opts, read_async_callback, read_async_info,
+                                io_handle, del_fn, nullptr /*dbg*/);
+  if (!s.ok()) {
+    delete read_async_info;
+  }
   return s;
 }
 
+void RandomAccessFileReader::ReadAsyncCallback(const FSReadRequest& req,
+                                               void* cb_arg) {
+  ReadAsyncInfo* read_async_info = static_cast<ReadAsyncInfo*>(cb_arg);
+  assert(read_async_info);
+  assert(read_async_info->cb_);
+
+  read_async_info->cb_(req, read_async_info->cb_arg_);
+
+  // Update stats and notify listeners.
+  if (stats_ != nullptr && file_read_hist_ != nullptr) {
+    // elapsed doesn't take into account delay and overwrite as StopWatch does
+    // in Read.
+    uint64_t elapsed = clock_->NowMicros() - read_async_info->start_time_;
+    file_read_hist_->Add(elapsed);
+  }
+  if (req.status.ok()) {
+    RecordInHistogram(stats_, ASYNC_READ_BYTES, req.result.size());
+  }
+#ifndef ROCKSDB_LITE
+  if (ShouldNotifyListeners()) {
+    auto finish_ts = FileOperationInfo::FinishNow();
+    NotifyOnFileReadFinish(req.offset, req.result.size(),
+                           read_async_info->fs_start_ts_, finish_ts,
+                           req.status);
+  }
+  if (!req.status.ok()) {
+    NotifyOnIOError(req.status, FileOperationType::kRead, file_name(),
+                    req.result.size(), req.offset);
+  }
+#endif
+  RecordIOStats(stats_, file_temperature_, is_last_level_, req.result.size());
+  delete read_async_info;
+}
 }  // namespace ROCKSDB_NAMESPACE

package/deps/rocksdb/rocksdb/file/random_access_file_reader.h

@@ -14,7 +14,6 @@
 
 #include "env/file_system_tracer.h"
 #include "port/port.h"
-#include "rocksdb/env.h"
 #include "rocksdb/file_system.h"
 #include "rocksdb/listener.h"
 #include "rocksdb/options.h"
@@ -24,6 +23,7 @@
 namespace ROCKSDB_NAMESPACE {
 class Statistics;
 class HistogramImpl;
+class SystemClock;
 
 using AlignedBuf = std::unique_ptr<char[]>;
 
@@ -38,7 +38,7 @@ FSReadRequest Align(const FSReadRequest& r, size_t alignment);
 // Otherwise, do nothing and return false.
 bool TryMerge(FSReadRequest* dest, const FSReadRequest& src);
 
-// RandomAccessFileReader is a wrapper on top of Env::RnadomAccessFile. It is
+// RandomAccessFileReader is a wrapper on top of Env::RandomAccessFile. It is
 // responsible for:
 // - Handling Buffered and Direct reads appropriately.
 // - Rate limiting compaction reads.
@@ -53,43 +53,75 @@ class RandomAccessFileReader {
       const FileOperationInfo::FinishTimePoint& finish_ts,
       const Status& status) const {
     FileOperationInfo info(FileOperationType::kRead, file_name_, start_ts,
-                           finish_ts, status);
+                           finish_ts, status, file_temperature_);
     info.offset = offset;
     info.length = length;
 
     for (auto& listener : listeners_) {
       listener->OnFileReadFinish(info);
     }
+    info.status.PermitUncheckedError();
   }
+
+  void NotifyOnIOError(const IOStatus& io_status, FileOperationType operation,
+                       const std::string& file_path, size_t length,
+                       uint64_t offset) const {
+    if (listeners_.empty()) {
+      return;
+    }
+    IOErrorInfo io_error_info(io_status, operation, file_path, length, offset);
+
+    for (auto& listener : listeners_) {
+      listener->OnIOError(io_error_info);
+    }
+    io_status.PermitUncheckedError();
+  }
+
 #endif  // ROCKSDB_LITE
 
   bool ShouldNotifyListeners() const { return !listeners_.empty(); }
 
   FSRandomAccessFilePtr file_;
   std::string file_name_;
-  Env* env_;
+  SystemClock* clock_;
   Statistics* stats_;
   uint32_t hist_type_;
   HistogramImpl* file_read_hist_;
   RateLimiter* rate_limiter_;
   std::vector<std::shared_ptr<EventListener>> listeners_;
+  const Temperature file_temperature_;
+  const bool is_last_level_;
+
+  struct ReadAsyncInfo {
+#ifndef ROCKSDB_LITE
+    FileOperationInfo::StartTimePoint fs_start_ts_;
+#endif
+    uint64_t start_time_;
+    std::function<void(const FSReadRequest&, void*)> cb_;
+    void* cb_arg_;
+  };
 
  public:
   explicit RandomAccessFileReader(
       std::unique_ptr<FSRandomAccessFile>&& raf, const std::string& _file_name,
-      Env* _env = nullptr, const std::shared_ptr<IOTracer>& io_tracer = nullptr,
+      SystemClock* clock = nullptr,
+      const std::shared_ptr<IOTracer>& io_tracer = nullptr,
       Statistics* stats = nullptr, uint32_t hist_type = 0,
       HistogramImpl* file_read_hist = nullptr,
       RateLimiter* rate_limiter = nullptr,
-      const std::vector<std::shared_ptr<EventListener>>& listeners = {})
-      : file_(std::move(raf), io_tracer),
+      const std::vector<std::shared_ptr<EventListener>>& listeners = {},
+      Temperature file_temperature = Temperature::kUnknown,
+      bool is_last_level = false)
+      : file_(std::move(raf), io_tracer, _file_name),
         file_name_(std::move(_file_name)),
-        env_(_env),
+        clock_(clock),
         stats_(stats),
         hist_type_(hist_type),
         file_read_hist_(file_read_hist),
         rate_limiter_(rate_limiter),
-        listeners_() {
+        listeners_(),
+        file_temperature_(file_temperature),
+        is_last_level_(is_last_level) {
 #ifndef ROCKSDB_LITE
     std::for_each(listeners.begin(), listeners.end(),
                   [this](const std::shared_ptr<EventListener>& e) {
@@ -102,6 +134,10 @@ class RandomAccessFileReader {
 #endif
   }
 
+  static IOStatus Create(const std::shared_ptr<FileSystem>& fs,
+                         const std::string& fname, const FileOptions& file_opts,
+                         std::unique_ptr<RandomAccessFileReader>* reader,
+                         IODebugContext* dbg);
   RandomAccessFileReader(const RandomAccessFileReader&) = delete;
   RandomAccessFileReader& operator=(const RandomAccessFileReader&) = delete;
 
@@ -115,19 +151,28 @@ class RandomAccessFileReader {
   // 2. Otherwise, scratch is not used and can be null, the aligned_buf owns
   // the internally allocated buffer on return, and the result refers to a
   // region in aligned_buf.
-  Status Read(const IOOptions& opts, uint64_t offset, size_t n, Slice* result,
-              char* scratch, AlignedBuf* aligned_buf,
-              bool for_compaction = false) const;
+  //
+  // `rate_limiter_priority` is used to charge the internal rate limiter when
+  // enabled. The special value `Env::IO_TOTAL` makes this operation bypass the
+  // rate limiter.
+  IOStatus Read(const IOOptions& opts, uint64_t offset, size_t n, Slice* result,
+                char* scratch, AlignedBuf* aligned_buf,
+                Env::IOPriority rate_limiter_priority) const;
 
   // REQUIRES:
   // num_reqs > 0, reqs do not overlap, and offsets in reqs are increasing.
   // In non-direct IO mode, aligned_buf should be null;
   // In direct IO mode, aligned_buf stores the aligned buffer allocated inside
   // MultiRead, the result Slices in reqs refer to aligned_buf.
-  Status MultiRead(const IOOptions& opts, FSReadRequest* reqs, size_t num_reqs,
-                   AlignedBuf* aligned_buf) const;
-
-  Status Prefetch(uint64_t offset, size_t n) const {
+  //
+  // `rate_limiter_priority` will be used to charge the internal rate limiter.
+  // It is not yet supported so the client must provide the special value
+  // `Env::IO_TOTAL` to bypass the rate limiter.
+  IOStatus MultiRead(const IOOptions& opts, FSReadRequest* reqs,
+                     size_t num_reqs, AlignedBuf* aligned_buf,
+                     Env::IOPriority rate_limiter_priority) const;
+
+  IOStatus Prefetch(uint64_t offset, size_t n) const {
     return file_->Prefetch(offset, n, IOOptions(), nullptr);
   }
 
@@ -137,6 +182,13 @@ class RandomAccessFileReader {
 
   bool use_direct_io() const { return file_->use_direct_io(); }
 
-  Env* env() const { return env_; }
+  IOStatus PrepareIOOptions(const ReadOptions& ro, IOOptions& opts);
+
+  IOStatus ReadAsync(FSReadRequest& req, const IOOptions& opts,
+                     std::function<void(const FSReadRequest&, void*)> cb,
+                     void* cb_arg, void** io_handle, IOHandleDeleter* del_fn,
+                     Env::IOPriority rate_limiter_priority);
+
+  void ReadAsyncCallback(const FSReadRequest& req, void* cb_arg);
 };
 }  // namespace ROCKSDB_NAMESPACE

package/deps/rocksdb/rocksdb/file/random_access_file_reader_test.cc

@@ -38,11 +38,12 @@ class RandomAccessFileReaderTest : public testing::Test {
   }
 
   void Read(const std::string& fname, const FileOptions& opts,
-                std::unique_ptr<RandomAccessFileReader>* reader) {
+            std::unique_ptr<RandomAccessFileReader>* reader) {
     std::string fpath = Path(fname);
     std::unique_ptr<FSRandomAccessFile> f;
     ASSERT_OK(fs_->NewRandomAccessFile(fpath, opts, &f, nullptr));
-    (*reader).reset(new RandomAccessFileReader(std::move(f), fpath, env_));
+    reader->reset(new RandomAccessFileReader(std::move(f), fpath,
+                                             env_->GetSystemClock().get()));
   }
 
   void AssertResult(const std::string& content,
@@ -84,9 +85,9 @@ TEST_F(RandomAccessFileReaderTest, ReadDirectIO) {
   size_t len = page_size / 3;
   Slice result;
   AlignedBuf buf;
-  for (bool for_compaction : {true, false}) {
+  for (Env::IOPriority rate_limiter_priority : {Env::IO_LOW, Env::IO_TOTAL}) {
     ASSERT_OK(r->Read(IOOptions(), offset, len, &result, nullptr, &buf,
-                      for_compaction));
+                      rate_limiter_priority));
     ASSERT_EQ(result.ToString(), content.substr(offset, len));
   }
 }
@@ -137,8 +138,8 @@ TEST_F(RandomAccessFileReaderTest, MultiReadDirectIO) {
     reqs.push_back(std::move(r0));
     reqs.push_back(std::move(r1));
     AlignedBuf aligned_buf;
-    ASSERT_OK(
-        r->MultiRead(IOOptions(), reqs.data(), reqs.size(), &aligned_buf));
+    ASSERT_OK(r->MultiRead(IOOptions(), reqs.data(), reqs.size(), &aligned_buf,
+                           Env::IO_TOTAL /* rate_limiter_priority */));
 
     AssertResult(content, reqs);
 
@@ -182,8 +183,8 @@ TEST_F(RandomAccessFileReaderTest, MultiReadDirectIO) {
     reqs.push_back(std::move(r1));
     reqs.push_back(std::move(r2));
     AlignedBuf aligned_buf;
-    ASSERT_OK(
-        r->MultiRead(IOOptions(), reqs.data(), reqs.size(), &aligned_buf));
+    ASSERT_OK(r->MultiRead(IOOptions(), reqs.data(), reqs.size(), &aligned_buf,
+                           Env::IO_TOTAL /* rate_limiter_priority */));
 
     AssertResult(content, reqs);
 
@@ -227,8 +228,8 @@ TEST_F(RandomAccessFileReaderTest, MultiReadDirectIO) {
     reqs.push_back(std::move(r1));
     reqs.push_back(std::move(r2));
     AlignedBuf aligned_buf;
-    ASSERT_OK(
-        r->MultiRead(IOOptions(), reqs.data(), reqs.size(), &aligned_buf));
+    ASSERT_OK(r->MultiRead(IOOptions(), reqs.data(), reqs.size(), &aligned_buf,
+                           Env::IO_TOTAL /* rate_limiter_priority */));
 
     AssertResult(content, reqs);
 
@@ -264,8 +265,8 @@ TEST_F(RandomAccessFileReaderTest, MultiReadDirectIO) {
     reqs.push_back(std::move(r0));
     reqs.push_back(std::move(r1));
     AlignedBuf aligned_buf;
-    ASSERT_OK(
-        r->MultiRead(IOOptions(), reqs.data(), reqs.size(), &aligned_buf));
+    ASSERT_OK(r->MultiRead(IOOptions(), reqs.data(), reqs.size(), &aligned_buf,
+                           Env::IO_TOTAL /* rate_limiter_priority */));
 
     AssertResult(content, reqs);
 

package/deps/rocksdb/rocksdb/file/read_write_util.cc

@@ -17,48 +17,13 @@ namespace ROCKSDB_NAMESPACE {
 IOStatus NewWritableFile(FileSystem* fs, const std::string& fname,
                          std::unique_ptr<FSWritableFile>* result,
                          const FileOptions& options) {
+  TEST_SYNC_POINT_CALLBACK("NewWritableFile::FileOptions.temperature",
+                           const_cast<Temperature*>(&options.temperature));
   IOStatus s = fs->NewWritableFile(fname, options, result, nullptr);
-  TEST_KILL_RANDOM("NewWritableFile:0", rocksdb_kill_odds * REDUCE_ODDS2);
+  TEST_KILL_RANDOM_WITH_WEIGHT("NewWritableFile:0", REDUCE_ODDS2);
   return s;
 }
 
-bool ReadOneLine(std::istringstream* iss, SequentialFileReader* seq_file_reader,
-                 std::string* output, bool* has_data, Status* result) {
-  const int kBufferSize = 8192;
-  char buffer[kBufferSize + 1];
-  Slice input_slice;
-
-  std::string line;
-  bool has_complete_line = false;
-  while (!has_complete_line) {
-    if (std::getline(*iss, line)) {
-      has_complete_line = !iss->eof();
-    } else {
-      has_complete_line = false;
-    }
-    if (!has_complete_line) {
-      // if we're not sure whether we have a complete line,
-      // further read from the file.
-      if (*has_data) {
-        *result = seq_file_reader->Read(kBufferSize, &input_slice, buffer);
-      }
-      if (input_slice.size() == 0) {
-        // meaning we have read all the data
-        *has_data = false;
-        break;
-      } else {
-        iss->str(line + input_slice.ToString());
-        // reset the internal state of iss so that we can keep reading it.
-        iss->clear();
-        *has_data = (input_slice.size() == kBufferSize);
-        continue;
-      }
-    }
-  }
-  *output = line;
-  return *has_data || has_complete_line;
-}
-
 #ifndef NDEBUG
 bool IsFileSectorAligned(const size_t off, size_t sector_size) {
   return off % sector_size == 0;