https://github.com/zyn0217 created 
https://github.com/llvm/llvm-project/pull/132669

Thanks to the example provided by @MagentaTreehouse, I realized the assertion I 
added didn't cover all valid cases like, when inheriting from a class template 
specialization, the source of a synthesized template parameter might be an 
implicit specialization, whose inner function template is thus living at depth 
0, for which we don’t want it to overflow too.

>From 27a42c2b5186b8bd5a71ca7e8bc574439d21e7fd Mon Sep 17 00:00:00 2001
From: Younan Zhang <zyn7...@gmail.com>
Date: Mon, 24 Mar 2025 12:43:17 +0800
Subject: [PATCH] [Clang] Fix the assertion condition after b8d1f3d6
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Thanks to the example provided by @MagentaTreehouse, I realized the
assertion I added didn't cover all valid cases like, when inheriting
from a class template specialization, the source of a synthesized
template parameter might be an implicit specialization, whose inner
function template is thus living at depth 0, for which we don’t want
it to overflow too.
---
 clang/lib/Sema/SemaTemplateDeductionGuide.cpp | 12 +++++----
 clang/test/SemaTemplate/deduction-guide.cpp   | 26 +++++++++++++++++++
 2 files changed, 33 insertions(+), 5 deletions(-)

diff --git a/clang/lib/Sema/SemaTemplateDeductionGuide.cpp 
b/clang/lib/Sema/SemaTemplateDeductionGuide.cpp
index 9cfdb7596b660..e20130a33f368 100644
--- a/clang/lib/Sema/SemaTemplateDeductionGuide.cpp
+++ b/clang/lib/Sema/SemaTemplateDeductionGuide.cpp
@@ -376,12 +376,14 @@ struct ConvertConstructorToDeductionGuideTransform {
         if (NestedPattern)
           Args.addOuterRetainedLevels(NestedPattern->getTemplateDepth());
         auto [Depth, Index] = getDepthAndIndex(Param);
-        // Depth can still be 0 if FTD belongs to an explicit class template
-        // specialization with an empty template parameter list. In that case,
-        // we don't want the NewDepth to overflow, and it should remain 0.
+        // Depth can be 0 if FTD belongs to a class template specialization 
with
+        // an empty template parameter list (i.e. either an explicit full
+        // specialization or an implicit specialization) In that case, we don't
+        // want the NewDepth to overflow, and it should remain 0.
         assert(Depth ||
-               cast<ClassTemplateSpecializationDecl>(FTD->getDeclContext())
-                   ->isExplicitSpecialization());
+               isa<ClassTemplateDecl *>(
+                   cast<ClassTemplateSpecializationDecl>(FTD->getDeclContext())
+                       ->getSpecializedTemplateOrPartial()));
         NamedDecl *NewParam = transformTemplateParameter(
             SemaRef, DC, Param, Args, Index + Depth1IndexAdjustment,
             Depth ? Depth - 1 : 0);
diff --git a/clang/test/SemaTemplate/deduction-guide.cpp 
b/clang/test/SemaTemplate/deduction-guide.cpp
index ecd152abebd74..95e740a026550 100644
--- a/clang/test/SemaTemplate/deduction-guide.cpp
+++ b/clang/test/SemaTemplate/deduction-guide.cpp
@@ -723,3 +723,29 @@ void test() { NewDeleteAllocator abc(42); } // 
expected-error {{no viable constr
 // CHECK-NEXT:  `-ParmVarDecl {{.+}} 'T'
 
 } // namespace GH128691
+
+namespace GH132616_DeductionGuide {
+
+template <class T> struct A {
+  template <class U>
+  A(U);
+};
+
+template <typename>
+struct B: A<int> {
+  using A::A;
+};
+
+template <class T>
+B(T) -> B<T>;
+
+B b(24);
+
+// CHECK-LABEL: Dumping GH132616_DeductionGuide::<deduction guide for B>:
+// CHECK-NEXT: FunctionTemplateDecl {{.+}} implicit <deduction guide for B>
+// CHECK-NEXT: |-TemplateTypeParmDecl {{.+}} typename depth 0 index 0
+// CHECK-NEXT: |-TemplateTypeParmDecl {{.+}} class depth 0 index 1 U
+// CHECK-NEXT: `-CXXDeductionGuideDecl {{.+}} implicit <deduction guide for B> 
'auto (U) -> B<type-parameter-0-0>'
+// CHECK-NEXT:  `-ParmVarDecl {{.+}} 'U'
+
+} // namespace GH132616_DeductionGuide

_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to