[scudo][standalone] Introduce the Quarantine

[cryptoad]

Summary

The Quarantine is used to hold chunks for a little while prior to
actually releasing them for potential reuse. The code is pretty much
the same as the sanitizer_common one, with additional shuffling of
the quarantine batches to decrease predictability of allocation
patterns when it is enabled.

[morehouse]

lib/scudo/standalone/quarantine.h

138 ↗

(On Diff #197568)

Does range-based loop work here?

157 ↗

(On Diff #197568)

Since uptr, probably want %zd for all.

181 ↗

(On Diff #197568)

What range of MaxCacheSize will Scudo allow? I notice you removed handling for MaxCacheSize == 0 in put below.

188 ↗

(On Diff #197568)

I think these can be initLinkerInitialized.

188 ↗

(On Diff #197568)

Right now it's a no-op, but I think we should call Cache.initLinkerInitialized here too in case a future change makes it not a no-op.

[cryptoad]

lib/scudo/standalone/quarantine.h

138 ↗

(On Diff #197568)

If you could please clarify, I am unsure as to which range you are referring to.

181 ↗

(On Diff #197568)

This is due to the fact that the Combined will skip a put and directly deallocate a chunk if:

• the global quarantine cache size is 0
• or the size of the chunk is greater than quarantine_max_chunk_size
• or the size of a chunk is 0 (as requested by user)

This is the current behavior of Scudo with quarantineOrDeallocateChunk, except that since we were using the sanitizer_common Quarantine, the put were doing an extraneous comparison.

As for range of values, they will likely be identical to the ones currently implemented in the non-standalone version.

188 ↗

(On Diff #197568)

Thanks, it turned out StaticMutex didn't have an initLinkerInitialized, so I added one.

[cryptoad]

Addressing Matt's comments:

• changing some initializers to initLinkerInitialized. This also required implementing a no-op version of this function for StaticSpinMutex
• changed a %d to %zd

[morehouse]

lib/scudo/standalone/quarantine.h

138 ↗

(On Diff #197568)

for (const QuarantineBatch &Batch : List) {
  ...
}

157 ↗

(On Diff #197568)

Also for memory overhead

[cryptoad]

lib/scudo/standalone/quarantine.h

138 ↗

(On Diff #197568)

It actually does work, thanks! Changed it.

157 ↗

(On Diff #197568)

Doh! Sorry I missed it. Also I forgot to clang-format post change.

[cryptoad]

Addressing Matt's new comments:

• changing another %d to %zd
• changing the iterating loop to a range loop
• clang-format'ing the source

compiler-rt/trunk/lib/scudo/standalone/CMakeLists.txt

set(SCUDO_HEADERS
   flags_parser.h
   fuchsia.h
   interface.h
   internal_defs.h
   linux.h
   list.h
   mutex.h
   platform.h
+  quarantine.h
   release.h
   report.h
   secondary.h
   size_class_map.h
   stats.h
   string_utils.h
   vector.h)
 

compiler-rt/trunk/lib/scudo/standalone/mutex.h

 
 #include "atomic_helpers.h"
 #include "common.h"
 
 namespace scudo {
 
 class StaticSpinMutex {
 public:
+  void initLinkerInitialized() {}
   void init() { atomic_store_relaxed(&State, 0); }
 
   void lock() {
     if (tryLock())
       return;
     lockSlow();
   }
 

compiler-rt/trunk/lib/scudo/standalone/quarantine.h

+//===-- quarantine.h --------------------------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef SCUDO_QUARANTINE_H_
+#define SCUDO_QUARANTINE_H_
+
+#include "list.h"
+#include "mutex.h"
+#include "string_utils.h"
+
+namespace scudo {
+
+struct QuarantineBatch {
+  // With the following count, a batch (and the header that protects it) occupy
+  // 4096 bytes on 32-bit platforms, and 8192 bytes on 64-bit.
+  static const u32 MaxCount = 1019;
+  QuarantineBatch *Next;
+  uptr Size;
+  u32 Count;
+  void *Batch[MaxCount];
+
+  void init(void *Ptr, uptr Size) {
+    Count = 1;
+    Batch[0] = Ptr;
+    this->Size = Size + sizeof(QuarantineBatch); // Account for the Batch Size.
+  }
+
+  // The total size of quarantined nodes recorded in this batch.
+  uptr getQuarantinedSize() const { return Size - sizeof(QuarantineBatch); }
+
+  void push_back(void *Ptr, uptr Size) {
+    DCHECK_LT(Count, MaxCount);
+    Batch[Count++] = Ptr;
+    this->Size += Size;
+  }
+
+  bool canMerge(const QuarantineBatch *const From) const {
+    return Count + From->Count <= MaxCount;
+  }
+
+  void merge(QuarantineBatch *const From) {
+    DCHECK_LE(Count + From->Count, MaxCount);
+    DCHECK_GE(Size, sizeof(QuarantineBatch));
+
+    for (uptr I = 0; I < From->Count; ++I)
+      Batch[Count + I] = From->Batch[I];
+    Count += From->Count;
+    Size += From->getQuarantinedSize();
+
+    From->Count = 0;
+    From->Size = sizeof(QuarantineBatch);
+  }
+
+  void shuffle(u32 State) { ::scudo::shuffle(Batch, Count, &State); }
+};
+
+COMPILER_CHECK(sizeof(QuarantineBatch) <= (1U << 13)); // 8Kb.
+
+// Per-thread cache of memory blocks.
+template <typename Callback> class QuarantineCache {
+public:
+  void initLinkerInitialized() {}
+  void init() {
+    memset(this, 0, sizeof(*this));
+    initLinkerInitialized();
+  }
+
+  // Total memory used, including internal accounting.
+  uptr getSize() const { return atomic_load_relaxed(&Size); }
+  // Memory used for internal accounting.
+  uptr getOverheadSize() const { return List.size() * sizeof(QuarantineBatch); }
+
+  void enqueue(Callback Cb, void *Ptr, uptr Size) {
+    if (List.empty() || List.back()->Count == QuarantineBatch::MaxCount) {
+      QuarantineBatch *B =
+          reinterpret_cast<QuarantineBatch *>(Cb.allocate(sizeof(*B)));
+      DCHECK(B);
+      B->init(Ptr, Size);
+      enqueueBatch(B);
+    } else {
+      List.back()->push_back(Ptr, Size);
+      addToSize(Size);
+    }
+  }
+
+  void transfer(QuarantineCache *From) {
+    List.append_back(&From->List);
+    addToSize(From->getSize());
+    atomic_store_relaxed(&From->Size, 0);
+  }
+
+  void enqueueBatch(QuarantineBatch *B) {
+    List.push_back(B);
+    addToSize(B->Size);
+  }
+
+  QuarantineBatch *dequeueBatch() {
+    if (List.empty())
+      return nullptr;
+    QuarantineBatch *B = List.front();
+    List.pop_front();
+    subFromSize(B->Size);
+    return B;
+  }
+
+  void mergeBatches(QuarantineCache *ToDeallocate) {
+    uptr ExtractedSize = 0;
+    QuarantineBatch *Current = List.front();
+    while (Current && Current->Next) {
+      if (Current->canMerge(Current->Next)) {
+        QuarantineBatch *Extracted = Current->Next;
+        // Move all the chunks into the current batch.
+        Current->merge(Extracted);
+        DCHECK_EQ(Extracted->Count, 0);
+        DCHECK_EQ(Extracted->Size, sizeof(QuarantineBatch));
+        // Remove the next batch From the list and account for its Size.
+        List.extract(Current, Extracted);
+        ExtractedSize += Extracted->Size;
+        // Add it to deallocation list.
+        ToDeallocate->enqueueBatch(Extracted);
+      } else {
+        Current = Current->Next;
+      }
+    }
+    subFromSize(ExtractedSize);
+  }
+
+  void printStats() const {
+    uptr BatchCount = 0;
+    uptr TotalOverheadBytes = 0;
+    uptr TotalBytes = 0;
+    uptr TotalQuarantineChunks = 0;
+    for (const QuarantineBatch &Batch : List) {
+      BatchCount++;
+      TotalBytes += Batch.Size;
+      TotalOverheadBytes += Batch.Size - Batch.getQuarantinedSize();
+      TotalQuarantineChunks += Batch.Count;
+    }
+    const uptr QuarantineChunksCapacity =
+        BatchCount * QuarantineBatch::MaxCount;
+    const uptr ChunksUsagePercent =
+        (QuarantineChunksCapacity == 0)
+            ? 0
+            : TotalQuarantineChunks * 100 / QuarantineChunksCapacity;
+    const uptr TotalQuarantinedBytes = TotalBytes - TotalOverheadBytes;
+    const uptr MemoryOverheadPercent =
+        (TotalQuarantinedBytes == 0)
+            ? 0
+            : TotalOverheadBytes * 100 / TotalQuarantinedBytes;
+    Printf("Global quarantine stats: batches: %zd; bytes: %zd (user: %zd); "
+           "chunks: %zd (capacity: %zd); %zd%% chunks used; %zd%% memory "
+           "overhead\n",
+           BatchCount, TotalBytes, TotalQuarantinedBytes, TotalQuarantineChunks,
+           QuarantineChunksCapacity, ChunksUsagePercent, MemoryOverheadPercent);
+  }
+
+private:
+  IntrusiveList<QuarantineBatch> List;
+  atomic_uptr Size;
+
+  void addToSize(uptr add) { atomic_store_relaxed(&Size, getSize() + add); }
+  void subFromSize(uptr sub) { atomic_store_relaxed(&Size, getSize() - sub); }
+};
+
+// The callback interface is:
+// void Callback::recycle(Node *Ptr);
+// void *Callback::allocate(uptr Size);
+// void Callback::deallocate(void *Ptr);
+template <typename Callback, typename Node> class GlobalQuarantine {
+public:
+  typedef QuarantineCache<Callback> CacheT;
+
+  void initLinkerInitialized(uptr Size, uptr CacheSize) {
+    // Thread local quarantine size can be zero only when global quarantine size
+    // is zero (it allows us to perform just one atomic read per put() call).
+    CHECK((Size == 0 && CacheSize == 0) || CacheSize != 0);
+
+    atomic_store_relaxed(&MaxSize, Size);
+    atomic_store_relaxed(&MinSize, Size / 10 * 9); // 90% of max size.
+    atomic_store_relaxed(&MaxCacheSize, CacheSize);
+
+    Cache.initLinkerInitialized();
+    CacheMutex.initLinkerInitialized();
+    RecyleMutex.initLinkerInitialized();
+  }
+  void init(uptr Size, uptr CacheSize) {
+    memset(this, 0, sizeof(*this));
+    initLinkerInitialized(Size, CacheSize);
+  }
+
+  uptr getMaxSize() const { return atomic_load_relaxed(&MaxSize); }
+  uptr getCacheSize() const { return atomic_load_relaxed(&MaxCacheSize); }
+
+  void put(CacheT *C, Callback Cb, Node *Ptr, uptr Size) {
+    C->enqueue(Cb, Ptr, Size);
+    if (C->getSize() > getCacheSize())
+      drain(C, Cb);
+  }
+
+  void NOINLINE drain(CacheT *C, Callback Cb) {
+    {
+      SpinMutexLock L(&CacheMutex);
+      Cache.transfer(C);
+    }
+    if (Cache.getSize() > getMaxSize() && RecyleMutex.tryLock())
+      recycle(atomic_load_relaxed(&MinSize), Cb);
+  }
+
+  void NOINLINE drainAndRecycle(CacheT *C, Callback Cb) {
+    {
+      SpinMutexLock L(&CacheMutex);
+      Cache.transfer(C);
+    }
+    RecyleMutex.lock();
+    recycle(0, Cb);
+  }
+
+  void printStats() const {
+    // It assumes that the world is stopped, just as the allocator's printStats.
+    Printf("Quarantine limits: global: %zdM; thread local: %zdK\n",
+           getMaxSize() >> 20, getCacheSize() >> 10);
+    Cache.printStats();
+  }
+
+private:
+  // Read-only data.
+  alignas(SCUDO_CACHE_LINE_SIZE) StaticSpinMutex CacheMutex;
+  CacheT Cache;
+  alignas(SCUDO_CACHE_LINE_SIZE) StaticSpinMutex RecyleMutex;
+  atomic_uptr MinSize;
+  atomic_uptr MaxSize;
+  alignas(SCUDO_CACHE_LINE_SIZE) atomic_uptr MaxCacheSize;
+
+  void NOINLINE recycle(uptr MinSize, Callback Cb) {
+    CacheT Tmp;
+    Tmp.init();
+    {
+      SpinMutexLock L(&CacheMutex);
+      // Go over the batches and merge partially filled ones to
+      // save some memory, otherwise batches themselves (since the memory used
+      // by them is counted against quarantine limit) can overcome the actual
+      // user's quarantined chunks, which diminishes the purpose of the
+      // quarantine.
+      const uptr CacheSize = Cache.getSize();
+      const uptr OverheadSize = Cache.getOverheadSize();
+      DCHECK_GE(CacheSize, OverheadSize);
+      // Do the merge only when overhead exceeds this predefined limit (might
+      // require some tuning). It saves us merge attempt when the batch list
+      // quarantine is unlikely to contain batches suitable for merge.
+      constexpr uptr OverheadThresholdPercents = 100;
+      if (CacheSize > OverheadSize &&
+          OverheadSize * (100 + OverheadThresholdPercents) >
+              CacheSize * OverheadThresholdPercents) {
+        Cache.mergeBatches(&Tmp);
+      }
+      // Extract enough chunks from the quarantine to get below the max
+      // quarantine size and leave some leeway for the newly quarantined chunks.
+      while (Cache.getSize() > MinSize)
+        Tmp.enqueueBatch(Cache.dequeueBatch());
+    }
+    RecyleMutex.unlock();
+    doRecycle(&Tmp, Cb);
+  }
+
+  void NOINLINE doRecycle(CacheT *C, Callback Cb) {
+    while (QuarantineBatch *B = C->dequeueBatch()) {
+      const u32 Seed = static_cast<u32>(
+          (reinterpret_cast<uptr>(B) ^ reinterpret_cast<uptr>(C)) >> 4);
+      B->shuffle(Seed);
+      constexpr uptr NumberOfPrefetch = 8UL;
+      CHECK(NumberOfPrefetch <= ARRAY_SIZE(B->Batch));
+      for (uptr I = 0; I < NumberOfPrefetch; I++)
+        PREFETCH(B->Batch[I]);
+      for (uptr I = 0, Count = B->Count; I < Count; I++) {
+        if (I + NumberOfPrefetch < Count)
+          PREFETCH(B->Batch[I + NumberOfPrefetch]);
+        Cb.recycle(reinterpret_cast<Node *>(B->Batch[I]));
+      }
+      Cb.deallocate(B);
+    }
+  }
+};
+
+} // namespace scudo
+
+#endif // SCUDO_QUARANTINE_H_

compiler-rt/trunk/lib/scudo/standalone/tests/CMakeLists.txt

 set(SCUDO_UNIT_TEST_SOURCES
   atomic_test.cc
   bytemap_test.cc
   checksum_test.cc
   flags_test.cc
   list_test.cc
   map_test.cc
   mutex_test.cc
+  quarantine_test.cc
   release_test.cc
   report_test.cc
   secondary_test.cc
   size_class_map_test.cc
   stats_test.cc
   strings_test.cc
   vector_test.cc
   scudo_unit_test_main.cc)
 
 add_scudo_unittest(ScudoUnitTest
   SOURCES ${SCUDO_UNIT_TEST_SOURCES})

compiler-rt/trunk/lib/scudo/standalone/tests/quarantine_test.cc

+//===-- quarantine_test.cc --------------------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "quarantine.h"
+
+#include "gtest/gtest.h"
+
+#include <stdlib.h>
+
+static void *FakePtr = reinterpret_cast<void *>(0xFA83FA83);
+static const scudo::uptr BlockSize = 8UL;
+static const scudo::uptr LargeBlockSize = 16384UL;
+
+struct QuarantineCallback {
+  void recycle(void *P) { EXPECT_EQ(P, FakePtr); }
+  void *allocate(scudo::uptr Size) { return malloc(Size); }
+  void deallocate(void *P) { free(P); }
+};
+
+typedef scudo::GlobalQuarantine<QuarantineCallback, void> QuarantineT;
+typedef typename QuarantineT::CacheT CacheT;
+
+static QuarantineCallback Cb;
+
+static void deallocateCache(CacheT *Cache) {
+  while (scudo::QuarantineBatch *Batch = Cache->dequeueBatch())
+    Cb.deallocate(Batch);
+}
+
+TEST(ScudoQuarantineTest, QuarantineBatchMerge) {
+  // Verify the trivial case.
+  scudo::QuarantineBatch Into;
+  Into.init(FakePtr, 4UL);
+  scudo::QuarantineBatch From;
+  From.init(FakePtr, 8UL);
+
+  Into.merge(&From);
+
+  EXPECT_EQ(Into.Count, 2UL);
+  EXPECT_EQ(Into.Batch[0], FakePtr);
+  EXPECT_EQ(Into.Batch[1], FakePtr);
+  EXPECT_EQ(Into.Size, 12UL + sizeof(scudo::QuarantineBatch));
+  EXPECT_EQ(Into.getQuarantinedSize(), 12UL);
+
+  EXPECT_EQ(From.Count, 0UL);
+  EXPECT_EQ(From.Size, sizeof(scudo::QuarantineBatch));
+  EXPECT_EQ(From.getQuarantinedSize(), 0UL);
+
+  // Merge the batch to the limit.
+  for (scudo::uptr I = 2; I < scudo::QuarantineBatch::MaxCount; ++I)
+    From.push_back(FakePtr, 8UL);
+  EXPECT_TRUE(Into.Count + From.Count == scudo::QuarantineBatch::MaxCount);
+  EXPECT_TRUE(Into.canMerge(&From));
+
+  Into.merge(&From);
+  EXPECT_TRUE(Into.Count == scudo::QuarantineBatch::MaxCount);
+
+  // No more space, not even for one element.
+  From.init(FakePtr, 8UL);
+
+  EXPECT_FALSE(Into.canMerge(&From));
+}
+
+TEST(ScudoQuarantineTest, QuarantineCacheMergeBatchesEmpty) {
+  CacheT Cache;
+  CacheT ToDeallocate;
+  Cache.init();
+  ToDeallocate.init();
+  Cache.mergeBatches(&ToDeallocate);
+
+  EXPECT_EQ(ToDeallocate.getSize(), 0UL);
+  EXPECT_EQ(ToDeallocate.dequeueBatch(), nullptr);
+}
+
+TEST(SanitizerCommon, QuarantineCacheMergeBatchesOneBatch) {
+  CacheT Cache;
+  Cache.init();
+  Cache.enqueue(Cb, FakePtr, BlockSize);
+  EXPECT_EQ(BlockSize + sizeof(scudo::QuarantineBatch), Cache.getSize());
+
+  CacheT ToDeallocate;
+  ToDeallocate.init();
+  Cache.mergeBatches(&ToDeallocate);
+
+  // Nothing to merge, nothing to deallocate.
+  EXPECT_EQ(BlockSize + sizeof(scudo::QuarantineBatch), Cache.getSize());
+
+  EXPECT_EQ(ToDeallocate.getSize(), 0UL);
+  EXPECT_EQ(ToDeallocate.dequeueBatch(), nullptr);
+
+  deallocateCache(&Cache);
+}
+
+TEST(ScudoQuarantineTest, QuarantineCacheMergeBatchesSmallBatches) {
+  // Make a Cache with two batches small enough to merge.
+  CacheT From;
+  From.init();
+  From.enqueue(Cb, FakePtr, BlockSize);
+  CacheT Cache;
+  Cache.init();
+  Cache.enqueue(Cb, FakePtr, BlockSize);
+
+  Cache.transfer(&From);
+  EXPECT_EQ(BlockSize * 2 + sizeof(scudo::QuarantineBatch) * 2,
+            Cache.getSize());
+
+  CacheT ToDeallocate;
+  ToDeallocate.init();
+  Cache.mergeBatches(&ToDeallocate);
+
+  // Batches merged, one batch to deallocate.
+  EXPECT_EQ(BlockSize * 2 + sizeof(scudo::QuarantineBatch), Cache.getSize());
+  EXPECT_EQ(ToDeallocate.getSize(), sizeof(scudo::QuarantineBatch));
+
+  deallocateCache(&Cache);
+  deallocateCache(&ToDeallocate);
+}
+
+TEST(ScudoQuarantineTest, QuarantineCacheMergeBatchesTooBigToMerge) {
+  const scudo::uptr NumBlocks = scudo::QuarantineBatch::MaxCount - 1;
+
+  // Make a Cache with two batches small enough to merge.
+  CacheT From;
+  CacheT Cache;
+  From.init();
+  Cache.init();
+  for (scudo::uptr I = 0; I < NumBlocks; ++I) {
+    From.enqueue(Cb, FakePtr, BlockSize);
+    Cache.enqueue(Cb, FakePtr, BlockSize);
+  }
+  Cache.transfer(&From);
+  EXPECT_EQ(BlockSize * NumBlocks * 2 + sizeof(scudo::QuarantineBatch) * 2,
+            Cache.getSize());
+
+  CacheT ToDeallocate;
+  ToDeallocate.init();
+  Cache.mergeBatches(&ToDeallocate);
+
+  // Batches cannot be merged.
+  EXPECT_EQ(BlockSize * NumBlocks * 2 + sizeof(scudo::QuarantineBatch) * 2,
+            Cache.getSize());
+  EXPECT_EQ(ToDeallocate.getSize(), 0UL);
+
+  deallocateCache(&Cache);
+}
+
+TEST(ScudoQuarantineTest, QuarantineCacheMergeBatchesALotOfBatches) {
+  const scudo::uptr NumBatchesAfterMerge = 3;
+  const scudo::uptr NumBlocks =
+      scudo::QuarantineBatch::MaxCount * NumBatchesAfterMerge;
+  const scudo::uptr NumBatchesBeforeMerge = NumBlocks;
+
+  // Make a Cache with many small batches.
+  CacheT Cache;
+  Cache.init();
+  for (scudo::uptr I = 0; I < NumBlocks; ++I) {
+    CacheT From;
+    From.init();
+    From.enqueue(Cb, FakePtr, BlockSize);
+    Cache.transfer(&From);
+  }
+
+  EXPECT_EQ(BlockSize * NumBlocks +
+                sizeof(scudo::QuarantineBatch) * NumBatchesBeforeMerge,
+            Cache.getSize());
+
+  CacheT ToDeallocate;
+  ToDeallocate.init();
+  Cache.mergeBatches(&ToDeallocate);
+
+  // All blocks should fit Into 3 batches.
+  EXPECT_EQ(BlockSize * NumBlocks +
+                sizeof(scudo::QuarantineBatch) * NumBatchesAfterMerge,
+            Cache.getSize());
+
+  EXPECT_EQ(ToDeallocate.getSize(),
+            sizeof(scudo::QuarantineBatch) *
+                (NumBatchesBeforeMerge - NumBatchesAfterMerge));
+
+  deallocateCache(&Cache);
+  deallocateCache(&ToDeallocate);
+}
+
+static const scudo::uptr MaxQuarantineSize = 1024UL << 10; // 1MB
+static const scudo::uptr MaxCacheSize = 256UL << 10;       // 256KB
+
+TEST(ScudoQuarantineTest, GlobalQuarantine) {
+  QuarantineT Quarantine;
+  CacheT Cache;
+  Cache.init();
+  Quarantine.init(MaxQuarantineSize, MaxCacheSize);
+  EXPECT_EQ(Quarantine.getMaxSize(), MaxQuarantineSize);
+  EXPECT_EQ(Quarantine.getCacheSize(), MaxCacheSize);
+
+  bool DrainOccurred = false;
+  scudo::uptr CacheSize = Cache.getSize();
+  EXPECT_EQ(Cache.getSize(), 0UL);
+  // We quarantine enough blocks that a drain has to occur. Verify this by
+  // looking for a decrease of the size of the cache.
+  for (scudo::uptr I = 0; I < 128UL; I++) {
+    Quarantine.put(&Cache, Cb, FakePtr, LargeBlockSize);
+    if (!DrainOccurred && Cache.getSize() < CacheSize)
+      DrainOccurred = true;
+    CacheSize = Cache.getSize();
+  }
+  EXPECT_TRUE(DrainOccurred);
+
+  Quarantine.drainAndRecycle(&Cache, Cb);
+  EXPECT_EQ(Cache.getSize(), 0UL);
+
+  Quarantine.printStats();
+}
+
+void *populateQuarantine(void *Param) {
+  CacheT Cache;
+  Cache.init();
+  QuarantineT *Quarantine = reinterpret_cast<QuarantineT *>(Param);
+  for (scudo::uptr I = 0; I < 128UL; I++)
+    Quarantine->put(&Cache, Cb, FakePtr, LargeBlockSize);
+  return 0;
+}
+
+TEST(ScudoQuarantineTest, ThreadedGlobalQuarantine) {
+  QuarantineT Quarantine;
+  Quarantine.init(MaxQuarantineSize, MaxCacheSize);
+
+  const scudo::uptr NumberOfThreads = 32U;
+  pthread_t T[NumberOfThreads];
+  for (scudo::uptr I = 0; I < NumberOfThreads; I++)
+    pthread_create(&T[I], 0, populateQuarantine, &Quarantine);
+  for (scudo::uptr I = 0; I < NumberOfThreads; I++)
+    pthread_join(T[I], 0);
+
+  Quarantine.printStats();
+}