diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -4338,9 +4338,11 @@
 
   typedef ProcessingContextState ParsingClassState;
   ParsingClassState PushParsingClass() {
+    ParsingClassDepth++;
     return DelayedDiagnostics.pushUndelayed();
   }
   void PopParsingClass(ParsingClassState state) {
+    ParsingClassDepth--;
     DelayedDiagnostics.popUndelayed(state);
   }
 
@@ -6385,7 +6387,7 @@
                                          SourceLocation RBrac,
                                          const ParsedAttributesView &AttrList);
   void ActOnFinishCXXMemberDecls();
-  void ActOnFinishCXXNonNestedClass(Decl *D);
+  void ActOnFinishCXXNonNestedClass();
 
   void ActOnReenterCXXMethodParameter(Scope *S, ParmVarDecl *Param);
   unsigned ActOnReenterTemplateScope(Scope *S, Decl *Template);
@@ -11547,6 +11549,8 @@
   SmallVector<CXXMethodDecl*, 4> DelayedDllExportMemberFunctions;
 
 private:
+  int ParsingClassDepth = 0;
+
   class SavePendingParsedClassStateRAII {
   public:
     SavePendingParsedClassStateRAII(Sema &S) : S(S) { swapSavedState(); }
@@ -11556,8 +11560,6 @@
              "there shouldn't be any pending delayed exception spec checks");
       assert(S.DelayedEquivalentExceptionSpecChecks.empty() &&
              "there shouldn't be any pending delayed exception spec checks");
-      assert(S.DelayedDllExportClasses.empty() &&
-             "there shouldn't be any pending delayed DLL export classes");
       swapSavedState();
     }
 
@@ -11567,14 +11569,12 @@
         SavedOverridingExceptionSpecChecks;
     decltype(DelayedEquivalentExceptionSpecChecks)
         SavedEquivalentExceptionSpecChecks;
-    decltype(DelayedDllExportClasses) SavedDllExportClasses;
 
     void swapSavedState() {
       SavedOverridingExceptionSpecChecks.swap(
           S.DelayedOverridingExceptionSpecChecks);
       SavedEquivalentExceptionSpecChecks.swap(
           S.DelayedEquivalentExceptionSpecChecks);
-      SavedDllExportClasses.swap(S.DelayedDllExportClasses);
     }
   };
 
diff --git a/clang/lib/Parse/ParseDeclCXX.cpp b/clang/lib/Parse/ParseDeclCXX.cpp
--- a/clang/lib/Parse/ParseDeclCXX.cpp
+++ b/clang/lib/Parse/ParseDeclCXX.cpp
@@ -3367,7 +3367,7 @@
 
     // We've finished parsing everything, including default argument
     // initializers.
-    Actions.ActOnFinishCXXNonNestedClass(TagDecl);
+    Actions.ActOnFinishCXXNonNestedClass();
   }
 
   if (TagDecl)
diff --git a/clang/lib/Sema/SemaDeclCXX.cpp b/clang/lib/Sema/SemaDeclCXX.cpp
--- a/clang/lib/Sema/SemaDeclCXX.cpp
+++ b/clang/lib/Sema/SemaDeclCXX.cpp
@@ -11829,7 +11829,7 @@
   }
 }
 
-void Sema::ActOnFinishCXXNonNestedClass(Decl *D) {
+void Sema::ActOnFinishCXXNonNestedClass() {
   referenceDLLExportedClassMethods();
 
   if (!DelayedDllExportMemberFunctions.empty()) {
diff --git a/clang/lib/Sema/SemaTemplateInstantiate.cpp b/clang/lib/Sema/SemaTemplateInstantiate.cpp
--- a/clang/lib/Sema/SemaTemplateInstantiate.cpp
+++ b/clang/lib/Sema/SemaTemplateInstantiate.cpp
@@ -2226,8 +2226,10 @@
   CheckCompletedCXXClass(Instantiation);
 
   // Default arguments are parsed, if not instantiated. We can go instantiate
-  // default arg exprs for default constructors if necessary now.
-  ActOnFinishCXXNonNestedClass(Instantiation);
+  // default arg exprs for default constructors if necessary now. Unless we're
+  // parsing a class, in which case wait until that's finished.
+  if (ParsingClassDepth == 0)
+    ActOnFinishCXXNonNestedClass();
 
   // Instantiate late parsed attributes, and attach them to their decls.
   // See Sema::InstantiateAttrs
diff --git a/clang/test/CodeGenCXX/dllexport.cpp b/clang/test/CodeGenCXX/dllexport.cpp
--- a/clang/test/CodeGenCXX/dllexport.cpp
+++ b/clang/test/CodeGenCXX/dllexport.cpp
@@ -860,6 +860,20 @@
 };
 // M32-DAG: define weak_odr dso_local dllexport x86_thiscallcc %"struct.InClassInits::PR40006"* @"??0PR40006@InClassInits@@QAE@XZ"
 
+namespace pr40006 {
+// Delay emitting the method also past the instantiation of Tmpl<Inner>, i.e.
+// until the top-level class Outer is completely finished.
+template<typename> struct Tmpl {};
+struct Outer {
+    struct Inner {
+        __declspec(dllexport) Inner() = default;
+        unsigned int x = 0;
+    };
+    Tmpl<Inner> y;
+};
+// M32-DAG: define weak_odr dso_local dllexport x86_thiscallcc %"struct.InClassInits::pr40006::Outer::Inner"* @"??0Inner@Outer@pr40006@InClassInits@@QAE@XZ"
+}
+
 // PR42857: Clang would try to emit the non-trivial explicitly defaulted
 // dllexport ctor twice when doing an explicit instantiation definition.
 struct Qux { Qux(); };