Parallel: only allow the first TaskGroup to run tasks parallelly

Summary:
Concurrent (e.g. nested) llvm::parallel::for_each() may lead to dead
locks. See PR35788 (fixed by rLLD322041) and PR41508 (fixed by D60757).

When parallel_for_each() is about to return, in ~Latch() called by
~TaskGroup(), a thread (in the default executor) may block in
Latch::sync() waiting for Count to become zero. If all threads in the
default executor are blocked, it is a dead lock.

To fix this, force serial execution if the current TaskGroup is not the
first one. For a nested llvm::parallel::for_each(), this parallelizes
the outermost loop and serializes inner loops.

Differential Revision: https://reviews.llvm.org/D61115

llvm-svn: 359182

include/llvm/Support/Parallel.h

@@ -73,8 +73,12 @@ public:
 
 class TaskGroup {
   Latch L;
+  bool Parallel;
 
 public:
+  TaskGroup();
+  ~TaskGroup();
+
   void spawn(std::function<void()> f);
 
   void sync() const { L.sync(); }

lib/Support/Parallel.cpp

@@ -17,7 +17,9 @@
 #include <stack>
 #include <thread>
 
-using namespace llvm;
+namespace llvm {
+namespace parallel {
+namespace detail {
 
 namespace {
 
@@ -118,11 +120,28 @@ Executor *Executor::getDefaultExecutor() {
 #endif
 }
 
-void parallel::detail::TaskGroup::spawn(std::function<void()> F) {
-  L.inc();
-  Executor::getDefaultExecutor()->add([&, F] {
+static std::atomic<int> TaskGroupInstances;
+
+// Latch::sync() called by the dtor may cause one thread to block. If is a dead
+// lock if all threads in the default executor are blocked. To prevent the dead
+// lock, only allow the first TaskGroup to run tasks parallelly. In the scenario
+// of nested parallel_for_each(), only the outermost one runs parallelly.
+TaskGroup::TaskGroup() : Parallel(TaskGroupInstances++ == 0) {}
+TaskGroup::~TaskGroup() { --TaskGroupInstances; }
+
+void TaskGroup::spawn(std::function<void()> F) {
+  if (Parallel) {
+    L.inc();
+    Executor::getDefaultExecutor()->add([&, F] {
+      F();
+      L.dec();
+    });
+  } else {
     F();
-    L.dec();
-  });
+  }
 }
+
+} // namespace detail
+} // namespace parallel
+} // namespace llvm
 #endif // LLVM_ENABLE_THREADS