[scudo] Scudo thread specific data refactor, part 1

Summary:
We are going through an overhaul of Scudo's TSD, to allow for new platforms
to be integrated more easily, and make the code more sound.

This first part is mostly renaming, preferring some shorter names, correcting
some comments. I removed `getPrng` and `getAllocatorCache` to directly access
the members, there was not really any benefit to them (and it was suggested by
Dmitry in D37590).

The only functional change is in `scudo_tls_android.cpp`: we enforce bounds to
the `NumberOfTSDs` and most of the logic in `getTSDAndLockSlow` is skipped if we
only have 1 TSD.

Reviewers: alekseyshl, dvyukov, kcc

Reviewed By: dvyukov

Subscribers: llvm-commits, srhines

Differential Revision: https://reviews.llvm.org/D38139

llvm-svn: 313987

8 files changed, 90 insertions, 100 deletions

diff --git a/compiler-rt/lib/scudo/scudo_allocator.cpp b/compiler-rt/lib/scudo/scudo_allocator.cpp
index 9d65b86832f..e490a469d87 100644
--- a/compiler-rt/lib/scudo/scudo_allocator.cpp
+++ b/compiler-rt/lib/scudo/scudo_allocator.cpp
@@ -250,19 +250,11 @@ struct QuarantineCallback {
 typedef Quarantine<QuarantineCallback, ScudoChunk> ScudoQuarantine;
 typedef ScudoQuarantine::Cache ScudoQuarantineCache;
 COMPILER_CHECK(sizeof(ScudoQuarantineCache) <=
-               sizeof(ScudoThreadContext::QuarantineCachePlaceHolder));
+               sizeof(ScudoTSD::QuarantineCachePlaceHolder));
 
-AllocatorCache *getAllocatorCache(ScudoThreadContext *ThreadContext) {
-  return &ThreadContext->Cache;
-}
-
-ScudoQuarantineCache *getQuarantineCache(ScudoThreadContext *ThreadContext) {
-  return reinterpret_cast<
-      ScudoQuarantineCache *>(ThreadContext->QuarantineCachePlaceHolder);
-}
-
-ScudoPrng *getPrng(ScudoThreadContext *ThreadContext) {
-  return &ThreadContext->Prng;
+ScudoQuarantineCache *getQuarantineCache(ScudoTSD *TSD) {
+  return reinterpret_cast<ScudoQuarantineCache *>(
+      TSD->QuarantineCachePlaceHolder);
 }
 
 struct ScudoAllocator {
@@ -381,12 +373,11 @@ struct ScudoAllocator {
     uptr AllocSize;
     if (FromPrimary) {
       AllocSize = AlignedSize;
-      ScudoThreadContext *ThreadContext = getThreadContextAndLock();
-      if (LIKELY(ThreadContext)) {
-        Salt = getPrng(ThreadContext)->getU8();
-        Ptr = BackendAllocator.allocatePrimary(getAllocatorCache(ThreadContext),
-                                               AllocSize);
-        ThreadContext->unlock();
+      ScudoTSD *TSD = getTSDAndLock();
+      if (LIKELY(TSD)) {
+        Salt = TSD->Prng.getU8();
+        Ptr = BackendAllocator.allocatePrimary(&TSD->Cache, AllocSize);
+        TSD->unlock();
       } else {
         SpinMutexLock l(&FallbackMutex);
         Salt = FallbackPrng.getU8();
@@ -454,11 +445,10 @@ struct ScudoAllocator {
       Chunk->eraseHeader();
       void *Ptr = Chunk->getAllocBeg(Header);
       if (Header->FromPrimary) {
-        ScudoThreadContext *ThreadContext = getThreadContextAndLock();
-        if (LIKELY(ThreadContext)) {
-          getBackendAllocator().deallocatePrimary(
-              getAllocatorCache(ThreadContext), Ptr);
-          ThreadContext->unlock();
+        ScudoTSD *TSD = getTSDAndLock();
+        if (LIKELY(TSD)) {
+          getBackendAllocator().deallocatePrimary(&TSD->Cache, Ptr);
+          TSD->unlock();
         } else {
           SpinMutexLock Lock(&FallbackMutex);
           getBackendAllocator().deallocatePrimary(&FallbackAllocatorCache, Ptr);
@@ -476,13 +466,12 @@ struct ScudoAllocator {
       UnpackedHeader NewHeader = *Header;
       NewHeader.State = ChunkQuarantine;
       Chunk->compareExchangeHeader(&NewHeader, Header);
-      ScudoThreadContext *ThreadContext = getThreadContextAndLock();
-      if (LIKELY(ThreadContext)) {
-        AllocatorQuarantine.Put(getQuarantineCache(ThreadContext),
-                                QuarantineCallback(
-                                    getAllocatorCache(ThreadContext)),
+      ScudoTSD *TSD = getTSDAndLock();
+      if (LIKELY(TSD)) {
+        AllocatorQuarantine.Put(getQuarantineCache(TSD),
+                                QuarantineCallback(&TSD->Cache),
                                 Chunk, EstimatedSize);
-        ThreadContext->unlock();
+        TSD->unlock();
       } else {
         SpinMutexLock l(&FallbackMutex);
         AllocatorQuarantine.Put(&FallbackQuarantineCache,
@@ -607,11 +596,10 @@ struct ScudoAllocator {
     return allocate(NMemB * Size, MinAlignment, FromMalloc, true);
   }
 
-  void commitBack(ScudoThreadContext *ThreadContext) {
-    AllocatorCache *Cache = getAllocatorCache(ThreadContext);
-    AllocatorQuarantine.Drain(getQuarantineCache(ThreadContext),
-                              QuarantineCallback(Cache));
-    BackendAllocator.destroyCache(Cache);
+  void commitBack(ScudoTSD *TSD) {
+    AllocatorQuarantine.Drain(getQuarantineCache(TSD),
+                              QuarantineCallback(&TSD->Cache));
+    BackendAllocator.destroyCache(&TSD->Cache);
   }
 
   uptr getStats(AllocatorStat StatType) {
@@ -637,13 +625,13 @@ static void initScudoInternal(const AllocatorOptions &Options) {
   Instance.init(Options);
 }
 
-void ScudoThreadContext::init() {
+void ScudoTSD::init() {
   getBackendAllocator().initCache(&Cache);
   Prng.init();
   memset(QuarantineCachePlaceHolder, 0, sizeof(QuarantineCachePlaceHolder));
 }
 
-void ScudoThreadContext::commitBack() {
+void ScudoTSD::commitBack() {
   Instance.commitBack(this);
 }
 
diff --git a/compiler-rt/lib/scudo/scudo_tls.h b/compiler-rt/lib/scudo/scudo_tls.h
index 4784f6a305c..a3992e264ec 100644
--- a/compiler-rt/lib/scudo/scudo_tls.h
+++ b/compiler-rt/lib/scudo/scudo_tls.h
@@ -28,7 +28,7 @@ namespace __scudo {
 #include "scudo_tls_context_android.inc"
 #include "scudo_tls_context_linux.inc"
 
-struct ALIGNED(64) ScudoThreadContext : public ScudoThreadContextPlatform {
+struct ALIGNED(64) ScudoTSD : public ScudoTSDPlatform {
   AllocatorCache Cache;
   ScudoPrng Prng;
   uptr QuarantineCachePlaceHolder[4];
@@ -38,7 +38,7 @@ struct ALIGNED(64) ScudoThreadContext : public ScudoThreadContextPlatform {
 
 void initThread(bool MinimalInit);
 
-// Platform specific dastpath functions definitions.
+// Platform specific fastpath functions definitions.
 #include "scudo_tls_android.inc"
 #include "scudo_tls_linux.inc"
 
diff --git a/compiler-rt/lib/scudo/scudo_tls_android.cpp b/compiler-rt/lib/scudo/scudo_tls_android.cpp
index c0ea417ab86..3f215a72f0b 100644
--- a/compiler-rt/lib/scudo/scudo_tls_android.cpp
+++ b/compiler-rt/lib/scudo/scudo_tls_android.cpp
@@ -24,9 +24,9 @@ namespace __scudo {
 static pthread_once_t GlobalInitialized = PTHREAD_ONCE_INIT;
 static pthread_key_t PThreadKey;
 
-static atomic_uint32_t ThreadContextCurrentIndex;
-static ScudoThreadContext *ThreadContexts;
-static uptr NumberOfContexts;
+static atomic_uint32_t CurrentIndex;
+static ScudoTSD *TSDs;
+static u32 NumberOfTSDs;
 
 // sysconf(_SC_NPROCESSORS_{CONF,ONLN}) cannot be used as they allocate memory.
 static uptr getNumberOfCPUs() {
@@ -42,52 +42,55 @@ static void initOnce() {
   // TODO(kostyak): remove and restrict to N and above.
   CHECK_EQ(pthread_key_create(&PThreadKey, NULL), 0);
   initScudo();
-  NumberOfContexts = getNumberOfCPUs();
-  ThreadContexts = reinterpret_cast<ScudoThreadContext *>(
-      MmapOrDie(sizeof(ScudoThreadContext) * NumberOfContexts, __func__));
-  for (uptr i = 0; i < NumberOfContexts; i++)
-    ThreadContexts[i].init();
+  NumberOfTSDs = getNumberOfCPUs();
+  if (NumberOfTSDs == 0)
+    NumberOfTSDs = 1;
+  if (NumberOfTSDs > 32)
+    NumberOfTSDs = 32;
+  TSDs = reinterpret_cast<ScudoTSD *>(
+      MmapOrDie(sizeof(ScudoTSD) * NumberOfTSDs, "ScudoTSDs"));
+  for (u32 i = 0; i < NumberOfTSDs; i++)
+    TSDs[i].init();
 }
 
 void initThread(bool MinimalInit) {
   pthread_once(&GlobalInitialized, initOnce);
   // Initial context assignment is done in a plain round-robin fashion.
-  u32 Index = atomic_fetch_add(&ThreadContextCurrentIndex, 1,
-                               memory_order_relaxed);
-  ScudoThreadContext *ThreadContext =
-      &ThreadContexts[Index % NumberOfContexts];
-  *get_android_tls_ptr() = reinterpret_cast<uptr>(ThreadContext);
+  u32 Index = atomic_fetch_add(&CurrentIndex, 1, memory_order_relaxed);
+  ScudoTSD *TSD = &TSDs[Index % NumberOfTSDs];
+  *get_android_tls_ptr() = reinterpret_cast<uptr>(TSD);
 }
 
-ScudoThreadContext *getThreadContextAndLockSlow() {
-  ScudoThreadContext *ThreadContext;
-  // Go through all the contexts and find the first unlocked one. 
-  for (u32 i = 0; i < NumberOfContexts; i++) {
-    ThreadContext = &ThreadContexts[i];
-    if (ThreadContext->tryLock()) {
-      *get_android_tls_ptr() = reinterpret_cast<uptr>(ThreadContext);
-      return ThreadContext;
+ScudoTSD *getTSDAndLockSlow() {
+  ScudoTSD *TSD;
+  if (NumberOfTSDs > 1) {
+    // Go through all the contexts and find the first unlocked one.
+    for (u32 i = 0; i < NumberOfTSDs; i++) {
+      TSD = &TSDs[i];
+      if (TSD->tryLock()) {
+        *get_android_tls_ptr() = reinterpret_cast<uptr>(TSD);
+        return TSD;
+      }
     }
-  }
-  // No luck, find the one with the lowest precedence, and slow lock it.
-  u64 Precedence = UINT64_MAX;
-  for (u32 i = 0; i < NumberOfContexts; i++) {
-    u64 SlowLockPrecedence = ThreadContexts[i].getSlowLockPrecedence();
-    if (SlowLockPrecedence && SlowLockPrecedence < Precedence) {
-      ThreadContext = &ThreadContexts[i];
-      Precedence = SlowLockPrecedence;
+    // No luck, find the one with the lowest Precedence, and slow lock it.
+    u64 LowestPrecedence = UINT64_MAX;
+    for (u32 i = 0; i < NumberOfTSDs; i++) {
+      u64 Precedence = TSDs[i].getPrecedence();
+      if (Precedence && Precedence < LowestPrecedence) {
+        TSD = &TSDs[i];
+        LowestPrecedence = Precedence;
+      }
+    }
+    if (LIKELY(LowestPrecedence != UINT64_MAX)) {
+      TSD->lock();
+      *get_android_tls_ptr() = reinterpret_cast<uptr>(TSD);
+      return TSD;
     }
   }
-  if (LIKELY(Precedence != UINT64_MAX)) {
-    ThreadContext->lock();
-    *get_android_tls_ptr() = reinterpret_cast<uptr>(ThreadContext);
-    return ThreadContext;
-  }
-  // Last resort (can this happen?), stick with the current one.
-  ThreadContext =
-      reinterpret_cast<ScudoThreadContext *>(*get_android_tls_ptr());
-  ThreadContext->lock();
-  return ThreadContext;
+  // Last resort, stick with the current one.
+  TSD = reinterpret_cast<ScudoTSD *>(*get_android_tls_ptr());
+  TSD->lock();
+  return TSD;
 }
 
 }  // namespace __scudo
diff --git a/compiler-rt/lib/scudo/scudo_tls_android.inc b/compiler-rt/lib/scudo/scudo_tls_android.inc
index 6b82e49f55a..9f3ef1a234a 100644
--- a/compiler-rt/lib/scudo/scudo_tls_android.inc
+++ b/compiler-rt/lib/scudo/scudo_tls_android.inc
@@ -26,17 +26,16 @@ ALWAYS_INLINE void initThreadMaybe(bool MinimalInit = false) {
   initThread(MinimalInit);
 }
 
-ScudoThreadContext *getThreadContextAndLockSlow();
+ScudoTSD *getTSDAndLockSlow();
 
-ALWAYS_INLINE ScudoThreadContext *getThreadContextAndLock() {
-  ScudoThreadContext *ThreadContext =
-      reinterpret_cast<ScudoThreadContext *>(*get_android_tls_ptr());
-  CHECK(ThreadContext);
+ALWAYS_INLINE ScudoTSD *getTSDAndLock() {
+  ScudoTSD *TSD = reinterpret_cast<ScudoTSD *>(*get_android_tls_ptr());
+  CHECK(TSD);
   // Try to lock the currently associated context.
-  if (ThreadContext->tryLock())
-    return ThreadContext;
+  if (TSD->tryLock())
+    return TSD;
   // If it failed, go the slow path.
-  return getThreadContextAndLockSlow();
+  return getTSDAndLockSlow();
 }
 
 #endif  // SANITIZER_LINUX && SANITIZER_ANDROID
diff --git a/compiler-rt/lib/scudo/scudo_tls_context_android.inc b/compiler-rt/lib/scudo/scudo_tls_context_android.inc
index f1951319d48..4787ec7b613 100644
--- a/compiler-rt/lib/scudo/scudo_tls_context_android.inc
+++ b/compiler-rt/lib/scudo/scudo_tls_context_android.inc
@@ -20,33 +20,33 @@
 
 #if SANITIZER_LINUX && SANITIZER_ANDROID
 
-struct ScudoThreadContextPlatform {
+struct ScudoTSDPlatform {
   INLINE bool tryLock() {
     if (Mutex.TryLock()) {
-      atomic_store_relaxed(&SlowLockPrecedence, 0);
+      atomic_store_relaxed(&Precedence, 0);
       return true;
     }
-    if (atomic_load_relaxed(&SlowLockPrecedence) == 0)
-      atomic_store_relaxed(&SlowLockPrecedence, NanoTime());
+    if (atomic_load_relaxed(&Precedence) == 0)
+      atomic_store_relaxed(&Precedence, NanoTime());
     return false;
   }
 
   INLINE void lock() {
     Mutex.Lock();
-    atomic_store_relaxed(&SlowLockPrecedence, 0);
+    atomic_store_relaxed(&Precedence, 0);
   }
 
   INLINE void unlock() {
     Mutex.Unlock();
   }
 
-  INLINE u64 getSlowLockPrecedence() {
-    return atomic_load_relaxed(&SlowLockPrecedence);
+  INLINE u64 getPrecedence() {
+    return atomic_load_relaxed(&Precedence);
   }
 
  private:
   StaticSpinMutex Mutex;
-  atomic_uint64_t SlowLockPrecedence;
+  atomic_uint64_t Precedence;
 };
 
 #endif  // SANITIZER_LINUX && SANITIZER_ANDROID
diff --git a/compiler-rt/lib/scudo/scudo_tls_context_linux.inc b/compiler-rt/lib/scudo/scudo_tls_context_linux.inc
index 8d292bdbc93..9a24256f2d7 100644
--- a/compiler-rt/lib/scudo/scudo_tls_context_linux.inc
+++ b/compiler-rt/lib/scudo/scudo_tls_context_linux.inc
@@ -20,7 +20,7 @@
 
 #if SANITIZER_LINUX && !SANITIZER_ANDROID
 
-struct ScudoThreadContextPlatform {
+struct ScudoTSDPlatform {
   ALWAYS_INLINE void unlock() {}
 };
 
diff --git a/compiler-rt/lib/scudo/scudo_tls_linux.cpp b/compiler-rt/lib/scudo/scudo_tls_linux.cpp
index f2592266d06..84553945797 100644
--- a/compiler-rt/lib/scudo/scudo_tls_linux.cpp
+++ b/compiler-rt/lib/scudo/scudo_tls_linux.cpp
@@ -28,7 +28,7 @@ static pthread_key_t PThreadKey;
 __attribute__((tls_model("initial-exec")))
 THREADLOCAL ThreadState ScudoThreadState = ThreadNotInitialized;
 __attribute__((tls_model("initial-exec")))
-THREADLOCAL ScudoThreadContext ThreadLocalContext;
+THREADLOCAL ScudoTSD TSD;
 
 static void teardownThread(void *Ptr) {
   uptr I = reinterpret_cast<uptr>(Ptr);
@@ -43,7 +43,7 @@ static void teardownThread(void *Ptr) {
                                    reinterpret_cast<void *>(I - 1)) == 0))
       return;
   }
-  ThreadLocalContext.commitBack();
+  TSD.commitBack();
   ScudoThreadState = ThreadTornDown;
 }
 
@@ -59,7 +59,7 @@ void initThread(bool MinimalInit) {
     return;
   CHECK_EQ(pthread_setspecific(PThreadKey, reinterpret_cast<void *>(
       GetPthreadDestructorIterations())), 0);
-  ThreadLocalContext.init();
+  TSD.init();
   ScudoThreadState = ThreadInitialized;
 }
 
diff --git a/compiler-rt/lib/scudo/scudo_tls_linux.inc b/compiler-rt/lib/scudo/scudo_tls_linux.inc
index 53b80448540..492807c5806 100644
--- a/compiler-rt/lib/scudo/scudo_tls_linux.inc
+++ b/compiler-rt/lib/scudo/scudo_tls_linux.inc
@@ -29,7 +29,7 @@ enum ThreadState : u8 {
 __attribute__((tls_model("initial-exec")))
 extern THREADLOCAL ThreadState ScudoThreadState;
 __attribute__((tls_model("initial-exec")))
-extern THREADLOCAL ScudoThreadContext ThreadLocalContext;
+extern THREADLOCAL ScudoTSD TSD;
 
 ALWAYS_INLINE void initThreadMaybe(bool MinimalInit = false) {
   if (LIKELY(ScudoThreadState != ThreadNotInitialized))
@@ -37,10 +37,10 @@ ALWAYS_INLINE void initThreadMaybe(bool MinimalInit = false) {
   initThread(MinimalInit);
 }
 
-ALWAYS_INLINE ScudoThreadContext *getThreadContextAndLock() {
+ALWAYS_INLINE ScudoTSD *getTSDAndLock() {
   if (UNLIKELY(ScudoThreadState != ThreadInitialized))
     return nullptr;
-  return &ThreadLocalContext;
+  return &TSD;
 }
 
 #endif  // SANITIZER_LINUX && !SANITIZER_ANDROID