Index: lib/asan/asan_interceptors.cc
===================================================================
--- lib/asan/asan_interceptors.cc
+++ lib/asan/asan_interceptors.cc
@@ -166,9 +166,13 @@
 #include "sanitizer_common/sanitizer_common_syscalls.inc"
 
 static thread_return_t THREAD_CALLING_CONV asan_thread_start(void *arg) {
-  AsanThread *t = (AsanThread*)arg;
+  atomic_uintptr_t *pt = reinterpret_cast<atomic_uintptr_t *>(arg);
+  AsanThread *t = 0;
+  while ((t = reinterpret_cast<AsanThread *>(
+              atomic_load(pt, memory_order_acquire))) == 0)
+    internal_sched_yield();
   SetCurrentThread(t);
-  return t->ThreadStart(GetTid());
+  return t->ThreadStart(GetTid(), pt);
 }
 
 #if ASAN_INTERCEPT_PTHREAD_CREATE
@@ -182,12 +186,19 @@
   int detached = 0;
   if (attr != 0)
     REAL(pthread_attr_getdetachstate)(attr, &detached);
-
-  u32 current_tid = GetCurrentTidOrInvalid();
-  AsanThread *t = AsanThread::Create(start_routine, arg);
-  CreateThreadContextArgs args = { t, &stack };
-  asanThreadRegistry().CreateThread(*(uptr*)t, detached, current_tid, &args);
-  return REAL(pthread_create)(thread, attr, asan_thread_start, t);
+  atomic_uintptr_t atomic_t;
+  atomic_store(&atomic_t, 0, memory_order_relaxed);
+  int result = REAL(pthread_create)(thread, attr, asan_thread_start, &atomic_t);
+  if (result == 0) {
+    u32 current_tid = GetCurrentTidOrInvalid();
+    AsanThread *t = AsanThread::Create(start_routine, arg);
+    CreateThreadContextArgs args = { t, &stack };
+    asanThreadRegistry().CreateThread(*(uptr*)t, detached, current_tid, &args);
+    atomic_store(&atomic_t, reinterpret_cast<uptr>(t), memory_order_release);
+    while (atomic_load(&atomic_t, memory_order_acquire) != 0)
+      internal_sched_yield();
+  }
+  return result;
 }
 #endif  // ASAN_INTERCEPT_PTHREAD_CREATE
 
Index: lib/asan/asan_rtl.cc
===================================================================
--- lib/asan/asan_rtl.cc
+++ lib/asan/asan_rtl.cc
@@ -680,7 +680,8 @@
       0, true, 0, &create_main_args);
   CHECK_EQ(0, main_tid);
   SetCurrentThread(main_thread);
-  main_thread->ThreadStart(internal_getpid());
+  main_thread->ThreadStart(internal_getpid(),
+                           /* signal_thread_is_registered */ 0);
   force_interface_symbols();  // no-op.
   SanitizerInitializeUnwinder();
 
Index: lib/asan/asan_thread.h
===================================================================
--- lib/asan/asan_thread.h
+++ lib/asan/asan_thread.h
@@ -60,7 +60,8 @@
   void Destroy();
 
   void Init();  // Should be called from the thread itself.
-  thread_return_t ThreadStart(uptr os_id);
+  thread_return_t ThreadStart(uptr os_id,
+                              atomic_uintptr_t *signal_thread_is_registered);
 
   uptr stack_top() { return stack_top_; }
   uptr stack_bottom() { return stack_bottom_; }
Index: lib/asan/asan_thread.cc
===================================================================
--- lib/asan/asan_thread.cc
+++ lib/asan/asan_thread.cc
@@ -155,9 +155,13 @@
   AsanPlatformThreadInit();
 }
 
-thread_return_t AsanThread::ThreadStart(uptr os_id) {
+thread_return_t AsanThread::ThreadStart(
+    uptr os_id, atomic_uintptr_t *signal_thread_is_registered) {
   Init();
   asanThreadRegistry().StartThread(tid(), os_id, 0);
+  if (signal_thread_is_registered)
+    atomic_store(signal_thread_is_registered, 0, memory_order_release);
+
   if (common_flags()->use_sigaltstack) SetAlternateSignalStack();
 
   if (!start_routine_) {
Index: lib/lsan/lsan_interceptors.cc
===================================================================
--- lib/lsan/lsan_interceptors.cc
+++ lib/lsan/lsan_interceptors.cc
@@ -215,9 +215,9 @@
   int tid = 0;
   while ((tid = atomic_load(&p->tid, memory_order_acquire)) == 0)
     internal_sched_yield();
-  atomic_store(&p->tid, 0, memory_order_release);
   SetCurrentThread(tid);
   ThreadStart(tid, GetTid());
+  atomic_store(&p->tid, 0, memory_order_release);
   return callback(param);
 }
 
Index: test/lsan/TestCases/leak_check_before_thread_started.cc
===================================================================
--- /dev/null
+++ test/lsan/TestCases/leak_check_before_thread_started.cc
@@ -0,0 +1,15 @@
+// Regression test for http://llvm.org/bugs/show_bug.cgi?id=21621
+// This test relies on timing between threads, so any failures will be flaky.
+// RUN: LSAN_BASE="use_stacks=0:use_registers=0"
+// RUN: %clangxx_lsan %s -std=c++11 -o %t
+// RUN: %run %t
+#include <thread>
+#include <chrono>
+
+void func() {
+      std::this_thread::sleep_for(std::chrono::milliseconds(500));
+}
+
+int main() {
+      std::thread(func).detach();
+}