[Clang] Correctly propagate type aliases' unexpanded flags up to lambda

We should have been checking desugar() for the type of the right-hand side
of a typedef declaration, instead of using getCanonicalType(), which
points to the end of the type alias chain.

Author: zyn0217

clang/docs/ReleaseNotes.rst

@@ -810,7 +810,7 @@ Bug Fixes to C++ Support
   module imports in those situations. (#GH60336)
 - Fix init-capture packs having a size of one before being instantiated. (#GH63677)
 - Clang now preserves the unexpanded flag in a lambda transform used for pack expansion. (#GH56852), (#GH85667),
-  (#GH99877).
+  (#GH99877), (#GH122417).
 - Fixed a bug when diagnosing ambiguous explicit specializations of constrained member functions.
 - Fixed an assertion failure when selecting a function from an overload set that includes a
   specialization of a conversion function template.

clang/lib/Sema/TreeTransform.h

@@ -8496,7 +8496,7 @@ TreeTransform<Derived>::TransformDeclStmt(DeclStmt *S) {
             getSema()
                 .getASTContext()
                 .getTypeDeclType(TD)
-                .getCanonicalType()
+                .getSingleStepDesugaredType(getSema().getASTContext())
                 ->containsUnexpandedParameterPack();
 
       if (auto *VD = dyn_cast<VarDecl>(Transformed))

clang/test/SemaCXX/fold_lambda_with_variadics.cpp

@@ -7,6 +7,8 @@ struct identity {
   using type = T;
 };
 
+template <class> using ElementType = int;
+
 template <class = void> void f() {
 
   static_assert([]<class... Is>(Is... x) {
@@ -47,6 +49,10 @@ template <class = void> void f() {
     }(), ...);
   }(1, 2);
 
+  []<class... Is>(Is...) {
+    ([] { using T = ElementType<Is>; }(), ...);
+  }(1);
+
   [](auto ...y) {
     ([y] { }(), ...);
   }();