[XRay][compiler-rt] Support in-memory processing of FDR mode logs

[dberris]

Summary

This change allows for handling the in-memory data associated with the
FDR mode implementation through the new __xray_log_process_buffers
API. With this change, we can now allow users to process the data
in-memory of the process instead of through writing files.

This for example allows users to stream the data of the FDR logging
implementation through network sockets, or through other mechanisms
instead of saving them to local files.

We introduce an FDR-specific flag, for "no_file_flush" which lets the
flushing logic skip opening/writing to files.

This option can be defaulted to true when building the compiler-rt
XRay runtime through the XRAY_FDR_OPTIONS preprocessor macro.

[dberris]

Rebase and minor change to ==/!= implementation.

[kpw]

compiler-rt/lib/xray/xray_buffer_queue.h

75 ↗

(On Diff #146276)

I have to ask. Why do you prefer this style where reference and pointer qualifiers are detached from the type they qualify? I understand the logic of grouping them with a variable name instead of a type, but choosing to group them to a function name baffles me.

97 ↗

(On Diff #146276)

Why DCHECK here instead of including this in the equality condition? I guess different views into the same Buffers are prohibited for now but this seems more fragile than it has to be.

compiler-rt/lib/xray/xray_fdr_logging.cc

74–76 ↗

(On Diff #146276)

Can you call this something like "nextBuffer" or "fdrNextBuffer" and write the comment so that it explains how it's supposed to be called? E.g. first call with a zero initialized XRayBuffer, then subsequent calls hand back the current XRayBuffer to advance forward.

Actually, since this maintains static iterators, you could keep static state about whether the header was returned and get rid of the parameter entirely if you choose. Although, you may need a way to reset the iterators, which is currently encoded in checks of the parameter.

Edit: It looks like this shouldn't be renamed since it's the function pointer installed by the __xray_log_set_buffer_iterator(). I still think explaining the usages and/or the assumptions about the state transitions is a good idea.

439 ↗

(On Diff #146276)

Why did you invert this condition?

compiler-rt/test/xray/TestCases/Posix/fdr-mode-inmemory.cc

1 ↗

(On Diff #146276)

In case any DCHECK failures are encountered during test execution, that should be covered by buildbots that build LLVM in debug mode and run tests right?

46 ↗

(On Diff #146276)

Clever FileCheck to just rely on asserts instead of CHECK statements.

[dberris]

• fixup: Address comments by kpw@.

compiler-rt/lib/xray/xray_buffer_queue.h

75 ↗

(On Diff #146276)

I don't -- clang-format does it for me. :)

97 ↗

(On Diff #146276)

This is a non-salient property of the type. This means, technically, the iterator which share the same base and offset compare equal to each other. Whether or not the bounds are the same don't determine whether two iterators are equal. The DCHECK is there precisely to only catch the case where the implementation has diverged from the intent, and only matters in cases where we want to catch that bug.

compiler-rt/lib/xray/xray_fdr_logging.cc

439 ↗

(On Diff #146276)

We change which flag(s) we're looking at in the fdr_impl header, which means we're consolidating the flags into the global fdrFlags() structure. This means we only change the FDR-specific implementation if there was no data in XRAY_FDR_OPTIONS, and revert to the data in the deprecated flags in XRAY_OPTIONS.

compiler-rt/test/xray/TestCases/Posix/fdr-mode-inmemory.cc

1 ↗

(On Diff #146276)

Yes.

lib/xray/xray_buffer_queue.h

 // This file is a part of XRay, a dynamic runtime instrumentation system.
 //
 // Defines the interface for a buffer queue implementation.
 //
 //===----------------------------------------------------------------------===//
 #ifndef XRAY_BUFFER_QUEUE_H
 #define XRAY_BUFFER_QUEUE_H
 
-#include <cstddef>
 #include "sanitizer_common/sanitizer_atomic.h"
+#include "sanitizer_common/sanitizer_common.h"
 #include "sanitizer_common/sanitizer_mutex.h"
+#include <cstddef>
 
 namespace __xray {
 
 /// BufferQueue implements a circular queue of fixed sized buffers (much like a
 /// freelist) but is concerned mostly with making it really quick to initialise,
 /// finalise, and get/return buffers to the queue. This is one key component of
 /// the "flight data recorder" (FDR) mode to support ongoing XRay function call
 /// trace collection.
 class BufferQueue {
- public:
   struct alignas(64) BufferExtents {
     __sanitizer::atomic_uint64_t Size;
   };
 
   struct Buffer {
     void *Data = nullptr;
     size_t Size = 0;
-    BufferExtents* Extents;
+    BufferExtents *Extents;
   };
 
- private:
   struct BufferRep {
     // The managed buffer.
     Buffer Buff;
 
     // This is true if the buffer has been returned to the available queue, and
     // is considered "used" by another thread.
     bool Used = false;
   };
 
+  // This models a ForwardIterator. |T| Must be either a `Buffer` or `const
+  // Buffer`. Note that we only advance to the "used" buffers, when
+  // incrementing, so that at dereference we're always at a valid point.
+  template <class T> class Iterator {
+  public:
+    BufferRep *Buffers = nullptr;
+    size_t Offset = 0;
+    size_t Max = 0;
+
+    Iterator &operator++() {
+      DCHECK_NE(Offset, Max);
+      do {
+        ++Offset;
+      } while (!Buffers[Offset].Used && Offset != Max);
+      return *this;
+    }
+
+    Iterator operator++(int) {
+      Iterator C = *this;
+      ++(*this);
+      return C;
+    }
+
+    T &operator*() const { return Buffers[Offset].Buff; }
+
+    T *operator->() const { return &(Buffers[Offset].Buff); }
+
+    Iterator(BufferRep *Root, size_t O, size_t M)
+        : Buffers(Root), Offset(O), Max(M) {
+      // We want to advance to the first Offset where the 'Used' property is
+      // true, or to the end of the list/queue.
+      while (!Buffers[Offset].Used && Offset != Max) {
+        ++Offset;
+      }
+    }
+
+    Iterator() = default;
+    Iterator(const Iterator &) = default;
+    Iterator(Iterator &&) = default;
+    Iterator &operator=(const Iterator &) = default;
+    Iterator &operator=(Iterator &&) = default;
+    ~Iterator() = default;
+
+    template <class V>
+    friend bool operator==(const Iterator &L, const Iterator<V> &R) {
+      DCHECK_EQ(L.Max, R.Max);
+      return L.Buffers == R.Buffers && L.Offset == R.Offset;
+    }
+
+    template <class V>
+    friend bool operator!=(const Iterator &L, const Iterator<V> &R) {
+      return !(L == R);
+    }
+  };
+
   // Size of each individual Buffer.
   size_t BufferSize;
 
   BufferRep *Buffers;
+
+  // Amount of pre-allocated buffers.
   size_t BufferCount;
 
   __sanitizer::SpinMutex Mutex;
   __sanitizer::atomic_uint8_t Finalizing;
 
   // Pointers to buffers managed/owned by the BufferQueue.
   void **OwnedBuffers;
 
   // Pointer to the next buffer to be handed out.
   BufferRep *Next;
 
   // Pointer to the entry in the array where the next released buffer will be
   // placed.
   BufferRep *First;
 
   // Count of buffers that have been handed out through 'getBuffer'.
   size_t LiveBuffers;
 
- public:
   enum class ErrorCode : unsigned {
     Ok,
     NotEnoughMemory,
     QueueFinalizing,
     UnrecognizedBuffer,
     AlreadyFinalized,
   };
 
   static const char *getErrorString(ErrorCode E) {
     switch (E) {
-      case ErrorCode::Ok:
-        return "(none)";
-      case ErrorCode::NotEnoughMemory:
-        return "no available buffers in the queue";
-      case ErrorCode::QueueFinalizing:
-        return "queue already finalizing";
-      case ErrorCode::UnrecognizedBuffer:
-        return "buffer being returned not owned by buffer queue";
-      case ErrorCode::AlreadyFinalized:
-        return "queue already finalized";
     }
     return "unknown error";
   }
 
   /// Initialise a queue of size |N| with buffers of size |B|. We report success
   /// through |Success|.
   BufferQueue(size_t B, size_t N, bool &Success);
 
public:
   ///
   /// After a call to finalize succeeds, all subsequent calls to finalize will
   /// fail with ErrorCode::QueueFinalizing.
   ErrorCode finalize();
 
   /// Applies the provided function F to each Buffer in the queue, only if the
   /// Buffer is marked 'used' (i.e. has been the result of getBuffer(...) and a
   /// releaseBuffer(...) operation).
-  template <class F>
-  void apply(F Fn) {
+  template <class F> void apply(F Fn) {
     __sanitizer::SpinMutexLock G(&Mutex);
-    for (auto I = Buffers, E = Buffers + BufferCount; I != E; ++I) {
-      const auto &T = *I;
-      if (T.Used) Fn(T.Buff);
-    }
+    for (auto I = begin(), E = end(); I != E; ++I)
+      Fn(*I);
+  }
+
+  using const_iterator = Iterator<const Buffer>;
+  using iterator = Iterator<Buffer>;
+
+  /// Provides iterator access to the raw Buffer instances.
+  iterator begin() const { return iterator(Buffers, 0, BufferCount); }
+  const_iterator cbegin() const {
+    return const_iterator(Buffers, 0, BufferCount);
+  iterator end() const { return iterator(Buffers, BufferCount, BufferCount); }
+  const_iterator cend() const {
+    return const_iterator(Buffers, BufferCount, BufferCount);
   }
 
   // Cleans up allocated buffers.
   ~BufferQueue();
 };
 
-}  // namespace __xray
+} // namespace __xray
 
-#endif  // XRAY_BUFFER_QUEUE_H
+#endif // XRAY_BUFFER_QUEUE_H

lib/xray/xray_fdr_flags.inc

XRAY_FLAG(int, func_duration_threshold_us, 5,
           "microseconds than this threshold.")
 XRAY_FLAG(int, grace_period_ms, 100,
           "FDR logging will wait this much time in milliseconds before "
           "actually flushing the log; this gives a chance for threads to "
           "notice that the log has been finalized and clean up.")
 XRAY_FLAG(int, buffer_size, 16384,
           "Size of buffers in the circular buffer queue.")
 XRAY_FLAG(int, buffer_max, 100, "Maximum number of buffers in the queue.")
+XRAY_FLAG(bool, no_file_flush, false,
+          "Set to true to not write log files by default.")

lib/xray/xray_fdr_logging.cc

 
 __sanitizer::atomic_sint32_t LogFlushStatus = {
     XRayLogFlushStatus::XRAY_LOG_NOT_FLUSHING};
 
 FDRLoggingOptions FDROptions;
 
 __sanitizer::SpinMutex FDROptionsMutex;
 
+namespace {
+XRayFileHeader &fdrCommonHeaderInfo() {
+  static XRayFileHeader Header = [] {
+    XRayFileHeader H;
+    // Version 2 of the log writes the extents of the buffer, instead of
+    // relying on an end-of-buffer record.
+    H.Version = 2;
+    H.Type = FileTypes::FDR_LOG;
+
+    // Test for required CPU features and cache the cycle frequency
+    static bool TSCSupported = probeRequiredCPUFeatures();
+    static uint64_t CycleFrequency =
+        TSCSupported ? getTSCFrequency() : __xray::NanosecondsPerSecond;
+    H.CycleFrequency = CycleFrequency;
+
+    // FIXME: Actually check whether we have 'constant_tsc' and
+    // 'nonstop_tsc' before setting the values in the header.
+    H.ConstantTSC = 1;
+    H.NonstopTSC = 1;
+    return H;
+  }();
+  return Header;
+}
+
+} // namespace
+
+// This is the iterator implementation, which knows how to handle FDR-mode
+// specific buffers. This is used as an implementation of the iterator function
+// needed by __xray_set_buffer_iterator(...). It maintains a global state of the
+// buffer iteration for the currently installed FDR mode buffers. In particular:
+//
+//   - If the argument represents the initial state of XRayBuffer ({nullptr, 0})
+//     then the iterator returns the header information.
+//   - If the argument represents the header information ({address of header
+//     info, size of the header info}) then it returns the first FDR buffer's
+//     address and extents.
+//   - It will keep returning the next buffer and extents as there are more
+//     buffers to process. When the input represents the last buffer, it will
+//     return the initial state to signal completion ({nullptr, 0}).
+//
+// See xray/xray_log_interface.h for more details on the requirements for the
+// implementations of __xray_set_buffer_iterator(...) and
+// __xray_log_process_buffers(...).
+XRayBuffer fdrIterator(const XRayBuffer B) {
+  DCHECK_EQ(__xray_current_mode(), "xray-fdr");
+  DCHECK(BQ->finalizing());
+
+  if (BQ == nullptr || !BQ->finalizing()) {
+    if (Verbosity())
+      Report(
+          "XRay FDR: Failed global buffer queue is null or not finalizing!\n");
+    return {nullptr, 0};
+  }
+
+  // We use a global scratch-pad for the header information, which only gets
+  // initialized the first time this function is called. We'll update one part
+  // of this information with some relevant data (in particular the number of
+  // buffers to expect).
+  static XRayFileHeader Header = fdrCommonHeaderInfo();
+  if (B.Data == nullptr && B.Size == 0) {
+    Header.FdrData = FdrAdditionalHeaderData{BQ->ConfiguredBufferSize()};
+    return XRayBuffer{static_cast<void *>(&Header), sizeof(Header)};
+  }
+
+  static BufferQueue::const_iterator It{};
+  static BufferQueue::const_iterator End{};
+  if (B.Data == static_cast<void *>(&Header) && B.Size == sizeof(Header)) {
+
+    // From this point on, we provide raw access to the raw buffer we're getting
+    // from the BufferQueue. We're relying on the iterators from the current
+    // Buffer queue.
+    It = BQ->cbegin();
+    End = BQ->cend();
+  }
+
+  if (It == End)
+    return {nullptr, 0};
+
+  XRayBuffer Result{
+      It->Data, __sanitizer::atomic_load(&It->Extents->Size,
+                                         __sanitizer::memory_order_acquire)};
+  ++It;
+  return Result;
+}
+
 // Must finalize before flushing.
 XRayLogFlushStatus fdrLoggingFlush() XRAY_NEVER_INSTRUMENT {
   if (__sanitizer::atomic_load(&LoggingStatus,
                                __sanitizer::memory_order_acquire) !=
       XRayLogInitStatus::XRAY_LOG_FINALIZED) {
     if (__sanitizer::Verbosity())
       Report("Not flushing log, implementation is not finalized.\n");
     return XRayLogFlushStatus::XRAY_LOG_NOT_FLUSHING;
   }
 
   s32 Result = XRayLogFlushStatus::XRAY_LOG_NOT_FLUSHING;
   if (!__sanitizer::atomic_compare_exchange_strong(
           &LogFlushStatus, &Result, XRayLogFlushStatus::XRAY_LOG_FLUSHING,
           __sanitizer::memory_order_release)) {
-
     if (__sanitizer::Verbosity())
       Report("Not flushing log, implementation is still finalizing.\n");
     return static_cast<XRayLogFlushStatus>(Result);
   }
 
   if (BQ == nullptr) {
     if (__sanitizer::Verbosity())
       Report("Cannot flush when global buffer queue is null.\n");
     return XRayLogFlushStatus::XRAY_LOG_NOT_FLUSHING;
   }
 
   // We wait a number of milliseconds to allow threads to see that we've
   // finalised before attempting to flush the log.
   __sanitizer::SleepForMillis(fdrFlags()->grace_period_ms);
 
+  // At this point, we're going to uninstall the iterator implementation, before
+  // we decide to do anything further with the global buffer queue.
+  __xray_log_remove_buffer_iterator();
+
+  if (fdrFlags()->no_file_flush) {
+    if (Verbosity())
+      Report("XRay FDR: Not flushing to file, 'no_file_flush=true'.\n");
+
+    // Clean up the buffer queue, and do not bother writing out the files!
+    delete BQ;
+    BQ = nullptr;
+    __sanitizer::atomic_store(&LogFlushStatus,
+                              XRayLogFlushStatus::XRAY_LOG_FLUSHED,
+                              __sanitizer::memory_order_release);
+    return XRayLogFlushStatus::XRAY_LOG_FLUSHED;
+  }
+
   // We write out the file in the following format:
   //
   //   1) We write down the XRay file header with version 1, type FDR_LOG.
   //   2) Then we use the 'apply' member of the BufferQueue that's live, to
   //      ensure that at this point in time we write down the buffers that have
   //      been released (and marked "used") -- we dump the full buffer for now
   //      (fixed-sized) and let the tools reading the buffers deal with the data
   //      afterwards.
   //
-  // FIXME: Support the case for letting users handle the data through
-  // __xray_process_buffers(...) and provide an option to skip writing files.
   int Fd = -1;
   {
     // FIXME: Remove this section of the code, when we remove the struct-based
     // configuration API.
     __sanitizer::SpinMutexLock Guard(&FDROptionsMutex);
     Fd = FDROptions.Fd;
   }
   if (Fd == -1)
     Fd = getLogFD();
   if (Fd == -1) {
     auto Result = XRayLogFlushStatus::XRAY_LOG_NOT_FLUSHING;
     __sanitizer::atomic_store(&LogFlushStatus, Result,
                               __sanitizer::memory_order_release);
     return Result;
   }
 
-  // Test for required CPU features and cache the cycle frequency
-  static bool TSCSupported = probeRequiredCPUFeatures();
-  static uint64_t CycleFrequency =
-      TSCSupported ? getTSCFrequency() : __xray::NanosecondsPerSecond;
-
-  XRayFileHeader Header;
-
-  // Version 2 of the log writes the extents of the buffer, instead of relying
-  // on an end-of-buffer record.
-  Header.Version = 2;
-  Header.Type = FileTypes::FDR_LOG;
-  Header.CycleFrequency = CycleFrequency;
-
-  // FIXME: Actually check whether we have 'constant_tsc' and 'nonstop_tsc'
-  // before setting the values in the header.
-  Header.ConstantTSC = 1;
-  Header.NonstopTSC = 1;
+  XRayFileHeader Header = fdrCommonHeaderInfo();
   Header.FdrData = FdrAdditionalHeaderData{BQ->ConfiguredBufferSize()};
   retryingWriteAll(Fd, reinterpret_cast<char *>(&Header),
                    reinterpret_cast<char *>(&Header) + sizeof(Header));
 
   BQ->apply([&](const BufferQueue::Buffer &B) {
     // Starting at version 2 of the FDR logging implementation, we only write
     // the records identified by the extents of the buffer. We use the Extents
-    // from the Buffer and write that out as the first record in the buffer.
-    // We still use a Metadata record, but fill in the extents instead for the
+    // from the Buffer and write that out as the first record in the buffer.  We
+    // still use a Metadata record, but fill in the extents instead for the
     // data.
     MetadataRecord ExtentsRecord;
     auto BufferExtents = __sanitizer::atomic_load(
         &B.Extents->Size, __sanitizer::memory_order_acquire);
     assert(BufferExtents <= B.Size);
     ExtentsRecord.Type = uint8_t(RecordType::Metadata);
     ExtentsRecord.RecordKind =
         uint8_t(MetadataRecord::RecordKinds::BufferExtents);

 
 struct TSCAndCPU {
   uint64_t TSC = 0;
   unsigned char CPU = 0;
 };
 
 static TSCAndCPU getTimestamp() XRAY_NEVER_INSTRUMENT {
   // We want to get the TSC as early as possible, so that we can check whether
-  // we've seen this CPU before. We also do it before we load anything else, to
-  // allow for forward progress with the scheduling.
+  // we've seen this CPU before. We also do it before we load anything else,
+  // to allow for forward progress with the scheduling.
   TSCAndCPU Result;
 
   // Test once for required CPU features
   static bool TSCSupported = probeRequiredCPUFeatures();
 
   if (TSCSupported) {
     Result.TSC = __xray::readTSC(Result.CPU);
   } else {
void fdrLoggingHandleCustomEvent(void *Event,
   }
   int32_t ReducedEventSize = static_cast<int32_t>(EventSize);
   auto &TLD = getThreadLocalData();
   if (!isLogInitializedAndReady(TLD.BQ, TSC, CPU, clock_gettime))
     return;
 
   // Here we need to prepare the log to handle:
   //   - The metadata record we're going to write. (16 bytes)
-  //   - The additional data we're going to write. Currently, that's the size of
-  //   the event we're going to dump into the log as free-form bytes.
+  //   - The additional data we're going to write. Currently, that's the size
+  //   of the event we're going to dump into the log as free-form bytes.
   if (!prepareBuffer(TSC, CPU, clock_gettime, MetadataRecSize + EventSize)) {
     TLD.BQ = nullptr;
     return;
   }
 
   // Write the custom event metadata record, which consists of the following
   // information:
   //   - 8 bytes (64-bits) for the full TSC when the event started.
void fdrLoggingHandleTypedEvent(
   }
   int32_t ReducedEventSize = static_cast<int32_t>(EventSize);
   auto &TLD = getThreadLocalData();
   if (!isLogInitializedAndReady(TLD.BQ, TSC, CPU, clock_gettime))
     return;
 
   // Here we need to prepare the log to handle:
   //   - The metadata record we're going to write. (16 bytes)
-  //   - The additional data we're going to write. Currently, that's the size of
-  //   the event we're going to dump into the log as free-form bytes.
+  //   - The additional data we're going to write. Currently, that's the size
+  //   of the event we're going to dump into the log as free-form bytes.
   if (!prepareBuffer(TSC, CPU, clock_gettime, MetadataRecSize + EventSize)) {
     TLD.BQ = nullptr;
     return;
   }
   // Write the custom event metadata record, which consists of the following
   // information:
   //   - 8 bytes (64-bits) for the full TSC when the event started.
   //   - 4 bytes (32-bits) for the length of the data.
if (BufferSize == 0 && BufferMax == 0) {
     // options until we migrate everyone to use the XRAY_FDR_OPTIONS
     // compiler-provided options.
     FDRParser.ParseString(useCompilerDefinedFlags());
     FDRParser.ParseString(useCompilerDefinedFDRFlags());
     auto *EnvOpts = GetEnv("XRAY_FDR_OPTIONS");
     if (EnvOpts == nullptr)
       EnvOpts = "";
     FDRParser.ParseString(EnvOpts);
+
+    // FIXME: Remove this when we fully remove the deprecated flags.
+    if (internal_strlen(EnvOpts) == 0) {
+      FDRFlags.func_duration_threshold_us =
+          flags()->xray_fdr_log_func_duration_threshold_us;
+      FDRFlags.grace_period_ms = flags()->xray_fdr_log_grace_period_ms;
+    }
+
+    // The provided options should always override the compiler-provided and
+    // environment-variable defined options.
     FDRParser.ParseString(static_cast<const char *>(Options));
     *fdrFlags() = FDRFlags;
-
     BufferSize = FDRFlags.buffer_size;
     BufferMax = FDRFlags.buffer_max;
-
-    // FIXME: Remove this when we fully remove the deprecated flags.
-    if (internal_strlen(EnvOpts) != 0) {
-      flags()->xray_fdr_log_func_duration_threshold_us =
-          FDRFlags.func_duration_threshold_us;
-      flags()->xray_fdr_log_grace_period_ms = FDRFlags.grace_period_ms;
-    }
+    __sanitizer::SpinMutexLock Guard(&FDROptionsMutex);
+    FDROptions.Fd = -1;
+    FDROptions.ReportErrors = true;
   } else if (OptionsSize != sizeof(FDRLoggingOptions)) {
     // FIXME: This is deprecated, and should really be removed.
     // At this point we use the flag parser specific to the FDR mode
     // implementation.
     if (__sanitizer::Verbosity())
       Report("Cannot initialize FDR logging; wrong size for options: %d\n",
              OptionsSize);
     return static_cast<XRayLogInitStatus>(__sanitizer::atomic_load(
XRayLogInitStatus fdrLoggingInit(size_t BufferSize, size_t BufferMax,
   // Arg1 handler should go in first to avoid concurrent code accidentally
   // falling back to arg0 when it should have ran arg1.
   __xray_set_handler_arg1(fdrLoggingHandleArg1);
   // Install the actual handleArg0 handler after initialising the buffers.
   __xray_set_handler(fdrLoggingHandleArg0);
   __xray_set_customevent_handler(fdrLoggingHandleCustomEvent);
   __xray_set_typedevent_handler(fdrLoggingHandleTypedEvent);
 
+  // Install the buffer iterator implementation.
+  __xray_log_set_buffer_iterator(fdrIterator);
+
   __sanitizer::atomic_store(&LoggingStatus,
                             XRayLogInitStatus::XRAY_LOG_INITIALIZED,
                             __sanitizer::memory_order_release);
 
   if (__sanitizer::Verbosity())
     Report("XRay FDR init successful.\n");
   return XRayLogInitStatus::XRAY_LOG_INITIALIZED;
 }

lib/xray/xray_fdr_logging_impl.h

 #include <time.h>
 #include <type_traits>
 #include <unistd.h>
 
 #include "sanitizer_common/sanitizer_common.h"
 #include "xray/xray_log_interface.h"
 #include "xray_buffer_queue.h"
 #include "xray_defs.h"
+#include "xray_fdr_flags.h"
 #include "xray_fdr_log_records.h"
-#include "xray_flags.h"
 #include "xray_tsc.h"
 
 namespace __xray {
 
 __sanitizer::atomic_sint32_t LoggingStatus = {
     XRayLogInitStatus::XRAY_LOG_UNINITIALIZED};
 
 /// We expose some of the state transitions when FDR logging mode is operating
writeFunctionRecord(int FuncId, uint32_t TSCDelta,
   incrementExtents(sizeof(FunctionRecord));
 }
 
 static uint64_t thresholdTicks() {
   static uint64_t TicksPerSec = probeRequiredCPUFeatures()
                                     ? getTSCFrequency()
                                     : __xray::NanosecondsPerSecond;
   static const uint64_t ThresholdTicks =
-      TicksPerSec * flags()->xray_fdr_log_func_duration_threshold_us / 1000000;
+      TicksPerSec * fdrFlags()->func_duration_threshold_us / 1000000;
   return ThresholdTicks;
 }
 
 // Re-point the thread local pointer into this thread's Buffer before the recent
 // "Function Entry" record and any "Tail Call Exit" records after that.
 static void rewindRecentCall(uint64_t TSC, uint64_t &LastTSC,
                              uint64_t &LastFunctionEntryTSC, int32_t FuncId) {
   auto &TLD = getThreadLocalData();

test/xray/TestCases/Posix/fdr-mode-inmemory.cc

+// RUN: %clangxx_xray -g -std=c++11 %s -o %t -fxray-modes=xray-fdr
+// RUN: rm fdr-inmemory-test-* || true
+// RUN: XRAY_OPTIONS="patch_premain=false xray_logfile_base=fdr-inmemory-test- \
+// RUN:     verbosity=1" \
+// RUN: XRAY_FDR_OPTIONS="no_file_flush=true func_duration_threshold_us=0" \
+// RUN:     %run %t 2>&1 | FileCheck %s
+// RUN: FILES=`find %T -name 'fdr-inmemory-test-*' | wc -l`
+// RUN: [ $FILES -eq 0 ]
+// RUN: rm fdr-inmemory-test-* || true
+//
+// REQUIRES: x86_64-target-arch
+// REQUIRES: built-in-llvm-tree
+
+#include "xray/xray_log_interface.h"
+#include <cassert>
+#include <iostream>
+
+uint64_t var = 0;
+uint64_t buffers = 0;
+[[clang::xray_always_instrument]] void __attribute__((noinline)) f() { ++var; }
+
+int main(int argc, char *argv[]) {
+  assert(__xray_log_select_mode("xray-fdr") ==
+         XRayLogRegisterStatus::XRAY_REGISTRATION_OK);
+  auto status = __xray_log_init_mode(
+      "xray-fdr",
+      "buffer_size=4096:buffer_max=10:func_duration_threshold_us=0");
+  assert(status == XRayLogInitStatus::XRAY_LOG_INITIALIZED);
+  __xray_patch();
+
+  // Create enough entries.
+  for (int i = 0; i != 1 << 20; ++i) {
+    f();
+  }
+
+  // Then we want to verify that we're getting 10 buffers outside of the initial
+  // header.
+  auto finalize_status = __xray_log_finalize();
+  assert(finalize_status == XRayLogInitStatus::XRAY_LOG_FINALIZED);
+  auto process_status =
+      __xray_log_process_buffers([](const char *, XRayBuffer) { ++buffers; });
+  std::cout << "buffers = " << buffers << std::endl;
+  assert(process_status == XRayLogFlushStatus::XRAY_LOG_FLUSHED);
+  auto flush_status = __xray_log_flushLog();
+  assert(flush_status == XRayLogFlushStatus::XRAY_LOG_FLUSHED);
+  // We expect 11 buffers because 1 header buffer + 10 actual FDR buffers.
+  // CHECK: Buffers = 11
+  std::cout << "Buffers = " << buffers << std::endl;
+  return 0;
+}