diff --git a/clang/lib/CodeGen/ItaniumCXXABI.cpp b/clang/lib/CodeGen/ItaniumCXXABI.cpp
index bd399b25c3f0..a12f08fbe9a3 100644
--- a/clang/lib/CodeGen/ItaniumCXXABI.cpp
+++ b/clang/lib/CodeGen/ItaniumCXXABI.cpp
@@ -2270,6 +2270,8 @@ void ItaniumCXXABI::EmitGuardedInit(CodeGenFunction &CGF,
 static void emitGlobalDtorWithCXAAtExit(CodeGenFunction &CGF,
                                         llvm::FunctionCallee dtor,
                                         llvm::Constant *addr, bool TLS) {
+  assert((TLS || CGF.getTypes().getCodeGenOpts().CXAAtExit) &&
+         "__cxa_atexit is disabled");
   const char *Name = "__cxa_atexit";
   if (TLS) {
     const llvm::Triple &T = CGF.getTarget().getTriple();
@@ -2363,13 +2365,13 @@ void ItaniumCXXABI::registerGlobalDtor(CodeGenFunction &CGF, const VarDecl &D,
   if (D.isNoDestroy(CGM.getContext()))
     return;
 
-  // Use __cxa_atexit if available.
-  if (CGM.getCodeGenOpts().CXAAtExit)
+  // emitGlobalDtorWithCXAAtExit will emit a call to either __cxa_thread_atexit
+  // or __cxa_atexit depending on whether this VarDecl is a thread-local storage
+  // or not. CXAAtExit controls only __cxa_atexit, so use it if it is enabled.
+  // We can always use __cxa_thread_atexit.
+  if (CGM.getCodeGenOpts().CXAAtExit || D.getTLSKind())
     return emitGlobalDtorWithCXAAtExit(CGF, dtor, addr, D.getTLSKind());
 
-  if (D.getTLSKind())
-    CGM.ErrorUnsupported(&D, "non-trivial TLS destruction");
-
   // In Apple kexts, we want to add a global destructor entry.
   // FIXME: shouldn't this be guarded by some variable?
   if (CGM.getLangOpts().AppleKext) {
diff --git a/clang/test/CodeGenCXX/cxx11-thread-local.cpp b/clang/test/CodeGenCXX/cxx11-thread-local.cpp
index de941af1afb8..d1590304be5b 100644
--- a/clang/test/CodeGenCXX/cxx11-thread-local.cpp
+++ b/clang/test/CodeGenCXX/cxx11-thread-local.cpp
@@ -4,6 +4,12 @@
 // RUN:     -triple x86_64-linux-gnu 2>&1 | FileCheck --check-prefix=CHECK --check-prefix=LINUX %s
 // RUN: %clang_cc1 -std=c++11 -emit-llvm %s -o - -triple x86_64-apple-darwin12 | FileCheck --check-prefix=CHECK --check-prefix=DARWIN %s
 
+// RUN: %clang_cc1 -std=c++11 -fno-use-cxa-atexit -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck --check-prefix=CHECK --check-prefix=LINUX %s
+// RUN: %clang_cc1 -std=c++11 -fno-use-cxa-atexit -emit-llvm %s -O2 -disable-llvm-passes -o - -triple x86_64-linux-gnu | FileCheck --check-prefix=CHECK --check-prefix=LINUX --check-prefix=CHECK-OPT %s
+// RUN: %clang_cc1 -std=c++11 -fno-use-cxa-atexit -femulated-tls -emit-llvm %s -o - \
+// RUN:     -triple x86_64-linux-gnu 2>&1 | FileCheck --check-prefix=CHECK --check-prefix=LINUX %s
+// RUN: %clang_cc1 -std=c++11 -fno-use-cxa-atexit -emit-llvm %s -o - -triple x86_64-apple-darwin12 | FileCheck --check-prefix=CHECK --check-prefix=DARWIN %s
+
 int f();
 int g();